Defining the genetic basis and evolutionary dynamics of adaptation is a central goal in evolutionary biology. Mutations underlying adaptation or biological innovation can depend on multiple factors including genetic backgrounds, phenotypic states, and genome architecture (Blount et al., 2008; Blount et al., 2012). One important class of mutation mediating adaptive evolution are copy number variants (CNVs) which comprise duplications or deletions of genomic sequences that range in size from gene fragments to whole chromosomes. Quantifying the rates at which CNVs occur, the factors that influence their formation, and the fitness and functional effects of CNVs is essential for understanding their role in evolutionary processes.

CNVs play roles in rapid adaptation in multiple contexts and are an initiating event in biological innovation. For example, in laboratory evolution experiments a spontaneous tandem duplication captured a promoter for expression of a citrate transporter and resulted in Escherichia coli cells, typically unable to use citrate, to start metabolizing citrate as a carbon source (Blount et al., 2012). CNVs can be beneficial in cancer cells, promote tumorigenesis (Ben-David and Amon, 2020), enhance cancer cell adaptability (Rutledge et al., 2016), and accelerate resistance to anti-cancer therapies (Lukow et al., 2021). Over longer time scales, CNVs serve as substrate from which new genes evolve (Ohno, 1970; Taylor and Raes, 2004) as duplicated genes redundant in function can accumulate mutations and evolve to acquire new functions. For example, the globin gene family in mammals arose from rounds of gene duplication and subsequent diversification (Storz, 2016). CNVs also contribute to macro-evolutionary processes and thereby contribute to species differences, such as between humans and chimpanzees (Cheng et al., 2005) and reproductive isolation (Zuellig and Sweigart, 2018).

Mutations, including CNVs, occur in part because of errors made during DNA replication or DNA repair. Two general processes underlie CNV formation: (1) DNA recombination-based mechanisms and (2) DNA replication-based mechanisms (Brewer et al., 2011; Harel et al., 2015; Hastings et al., 2009a; Malhotra and Sebat, 2012; Pös et al., 2021; Zhang et al., 2009a). Recombination-mediated mechanisms of CNV formation include non-allelic homologous recombination (NAHR) and nonhomologous end joining. NAHR occurs via recombination between homologous sequences that are not allelic. As such, NAHR occurs more frequently with repetitive sequences due to improper alignment of DNA segments and can occur either between (interchromosomal) or within (intrachromosomal) a chromosome (Harel et al., 2015). One prevalent class of repetitive sequence are retrotransposons and both full length and partial sequences, such as long terminal repeats (LTR), are substrates for homologous recombination generating gene amplifications (Avecilla et al., 2023; Dunham et al., 2002; Gresham et al., 2008; Lauer et al., 2018; Spealman et al., 2022). Separately, repetitive or palindromic DNA can form hairpin structures and thus are thought to be sites for CNV formation after double-strand DNA breakage (Lobachev et al., 2002; Narayanan et al., 2006). DNA replication-based mechanisms include fork stalling template switching (FoSTeS) and microhomology mediated break-induced replication (MMBIR; Carvalho et al., 2013; Gu et al., 2008; Hastings et al., 2009b; Lee et al., 2007). During FoSTeS and MMBIR, after a DNA replication fork stalls and a DNA replication error occurs in which the lagging strand switches to an incorrect template strand mediated by microhomology. Reinitiation of DNA synthesis at the incorrect site can form CNVs. A particular type of DNA replication-based error is Origin Dependent Inverted Repeat Amplification (ODIRA), in which short inverted repeats and an active origin of DNA replication enable template switching of the leading strand to the lagging strand template. Subsequent replication generates an intermediate DNA molecule that can recombine into the original genome to form a triplication with an inverted middle copy (Brewer et al., 2011; Brewer et al., 2015; Martin et al., 2024).

In microbes, CNVs can mediate rapid adaptation to selective conditions imposed through nutrient limitation in a chemostat. Selected CNVs often include genes encoding nutrient transporters that facilitate import of the limiting nutrient (Dunham et al., 2002; Gresham et al., 2008; Hong and Gresham, 2014; Horiuchi et al., 1963; Payen et al., 2016; Sonti and Roth, 1989), likely as a result of improved nutrient transport capacity due to increased protein production. Previous studies have found amplification of the general amino acid permease gene, GAP1, when Saccharomyces cerevisiae populations are continuously cultured in glutamine-limited chemostats (Gresham et al., 2010; Lauer et al., 2018). Amplification of GAP1 confers increased fitness in the selective environment (Avecilla et al., 2023; Lauer et al., 2018). Sequence characterization of these CNVs revealed that a diversity of de novo CNV alleles are generated and selected including tandem duplications, complex large CNVs, aneuploidies, and translocations. However, little is known about the molecular mechanisms underlying this diversity.

Local genome sequence elements are likely to be an important determinant of CNV formation rates and mechanisms. Genomic context can influence multiple properties including mutation rate, epigenetic regulation, chromatin state, transcription levels, DNA replication, and recombination rate (Arndt et al., 2005; Chuang and Li, 2004; Lang and Murray, 2011; Lercher and Hurst, 2002; Matassi et al., 1999; Nishant et al., 2009; Wolfe et al., 1989). Prior work has shown that CNVs occur more frequently in repetitive regions in the genome (Harel et al., 2015; Pentao et al., 1992; Stankiewicz et al., 2003; Turner et al., 2008). However, little is known about the role of local genomic architecture and organization on CNV formation rates, the types of CNVs that are generated, their associated fitness effects, and ultimately the paths taken during adaptive evolution.

Here, we aimed to investigate the effect of local genome architecture elements on de novo GAP1 CNV formation and selection dynamics during adaptive evolution of Saccharomyces cerevisiae. We hypothesized that sequence elements proximate to GAP1 potentiate CNV formation. The GAP1 locus, which is located on the short arm of chromosome XI, consists of two flanking Ty1 long terminal repeats (LTRs) that share 82% sequence identity and an origin of DNA replication or autonomously replicating sequence (ARS; Figure 1A). Both LTRs and ARS may facilitate GAP1 CNV formation due to their proximity. First, the flanking LTRs can undergo inter-chromatid NAHR to form tandem duplications of GAP1 on a linear chromosome (Lauer et al., 2018; Spealman et al., 2022). Second, intra-chromatid NAHR between the flanking LTRs can form an extrachromosomal circle containing GAP1 and an ARS able to self-propagate and integrate into the genome (Gresham et al., 2010). Finally, GAP1 triplications can form through ODIRA using short inverted repeats and the proximate ARS (Brewer et al., 2015; Lauer et al., 2018; Martin et al., 2024). These elements are thought to facilitate a high rate of GAP1 amplification, estimated to be on the order of 10^–4 per haploid genome per generation (Avecilla et al., 2022). To test our hypothesis we used a CNV reporter, wherein a constitutively expressed fluorescent GFP gene is inserted adjacent to GAP1 (Lauer et al., 2018). We engineered strains that lacked either the ARS (ARS∆), both flanking LTRs (LTR∆), or all three elements (ALL∆; Figure 1A). We performed experimental evolution using wildtype (WT) and genomic architecture mutant populations in glutamine-limited chemostats for 137 generations and quantified GAP1 CNVs using flow cytometry (Figure 1). Surprisingly, we find that the proximate DNA elements are not required for GAP1 CNV formation as GAP1 CNVs were identified in all evolving populations. We used neural network simulation-based inference (nnSBI) to infer the CNV formation rate and selection coefficient (Avecilla et al., 2022). We find that although genomic architecture mutants have significantly reduced CNV formation rates relative to WT and significantly lower selection coefficients, GAP1 CNVs repeatedly form and sweep to high frequency in all strains with modified genomes. We performed genome sequence analysis to define the molecular mechanisms of CNV formation for 177 CNV lineages and found that 26–80% of GAP1 CNVs are mediated by ODIRA across all four background strains. In the absence of the local ARS, a distal ARS facilitates CNV formation through ODIRA. We also find that homologous recombination mechanisms still mediate gene amplification in the absence of LTRs in part initiated by de novo insertion of retrotransposon elements at the locus. Our study reveals the remarkable plasticity of the genome and that template switching during DNA replication is a common source of adaptive CNVs even in the absence of the wildtype local DNA sequences.

Figure 1

Download asset Open asset

Accurate estimation of CNV allele frequencies in heterogeneous populations remains challenging using molecular methods such as DNA sequencing and qPCR. To address this challenge we previously developed a CNV reporter comprising a constitutively expressed fluorescent gene inserted upstream of GAP1 and observed recurrent amplification and selection of GAP1 in glutamine-limiting chemostats (Lauer et al., 2018). Subsequently, we showed that a high rate of GAP1 CNV formation and strong fitness effects explain the highly reproducible evolutionary dynamics (Avecilla et al., 2022). Noncoding sequence elements proximate to GAP1, including flanking LTRs in tandem orientation and an ARS, contribute to GAP1 CNV formation (Gresham et al., 2010; Lauer et al., 2018). Many studies have shown that repetitive sequence regions and origins of replications are hotspots of CNVs (Arlt et al., 2012; Cardoso et al., 2016; Di Rienzi et al., 2009; Gresham et al., 2010; Lauer et al., 2018; Zhang et al., 2013). Thus, we hypothesized that the local genomic architecture of GAP1 facilitates its high rate of CNV formation.

To test the role of proximate genomic features we engineered strains in which each element is deleted and thus differ from the wildtype strain (WT) containing a GAP1 CNV reporter by a single modification. Specifically, we constructed ARS∆, a strain lacking the single ARS, LTR∆, a strain lacking the flanking LTRs, and ALL∆, a strain lacking all three elements (Figure 1A). All strains contain the CNV reporter at the identical location as the WT strain. We confirmed scarless deletions of genetic elements using Sanger and whole-genome sequencing.

GAP1 amplifications can occur without CNV reporter amplification

In both WT and LTR∆ populations we observed that GAP1 CNV abundance stabilized around 75% during the equilibrium phase (Figure 1B) across each of the 12 independent populations (Figure 3—figure supplement 1). Flow cytometry analysis showed that each experiment begins with a population of cells with only one-copy of GFP (Figure 2A). Over generations, distinct populations appear with higher GFP fluorescence (Figure 2A). Previously, GFP fluorescence has been shown to scale with GAP1 copy number (Lauer et al., 2018). Therefore, the four distinct subpopulations observed (Figure 2A) likely represent cells harboring 1, 2, 3, and 4-copies of GAP1, respectively. This corroborates previous experimental evolution results in which de novo GAP1 CNVs are quickly formed and selected for and over selection higher copy outcompete lower copy subpopulations (Lauer et al., 2018). The raw flow cytometry plots (Figure 2—source data 1) and population GFP histograms (Figure 2—source data 2) also revealed a persistent single-copy GFP subpopulation throughout the timecourse (Figure 2A, bottom subpopulation in each panel). These data could be explained by two possible scenarios: (1) the existence of a non-GAP1 CNV subpopulation comprising beneficial variation at other loci with fitness effects equivalent to GAP1 CNVs; or (2) lineages with GAP1 CNVs without co-amplification of the CNV reporter. To resolve these two possibilities, we sequenced clones from the single-copy GFP subpopulation across the five WT populations from different chemostats (Supplementary file 1) and identified the presence of GAP1 amplifications without co-amplification of the CNV reporter in four out of five WT populations (Figure 2—figure supplement 1). We found eleven distinct GAP1 CNVs that lacked amplification of the reporter gene (Figure 2—figure supplement 1) indicating at least eleven independent CNV events occurred, either in the founder population or shortly after chemostat inoculation. By contrast, in one of the five populations, population 3, all clones from the single-copy GFP subpopulation contained one copy of GFP and one copy of GAP1 (Supplementary file 1), suggesting these clones have a beneficial mutation elsewhere in the genome that allows their stable coexistence with the GAP1 CNV subpopulation. The GAP1 CNVs without GFP amplification were either pre-existing at the time of the inoculation or occurred shortly after inoculation. We suspect that they likely occurred after inoculation but early in the evolution, for three reasons: (1) the similar ~75% plateau is observed in the dynamics in all independent WT and LTR∆ populations, (2) at least 1 independent CNVs of this type were detected (GAP1 amplification without co-amplification of the GFP), and (3) there are no common CNVs detected across chemostats. Our findings show that GAP1 amplification without coamplification of the CNV reporter can occur and beneficial variation other than GAP1 CNVs underlie adaptation to glutamine-limitation.

Figure 2 with 1 supplement see all

Download asset Open asset

Figure 2—source data 1

Raw flow cytometry plots over the long term experimental evolution.

FSC-A is forward scatter-area which is a proxy for cell size. GFP fluorescence was measured using the B2-A channel in arbitrary units. Hierarchical gating was performed to identify zero-, one-, and two-or-more-copy populations. Within the two-or-more copy gate, distinct subpopulations formed consistent with having a two-, three-, four- copies of GFP.

https://cdn.elifesciences.org/articles/98934/elife-98934-fig2-data1-v1.pdf

Download elife-98934-fig2-data1-v1.pdf

Figure 2—source data 2

Population GFP Ridgeplots.

Density plots of cell-size normalized GFP fluorescence in arbitrary units (a.u.) for every population and timepoint over the course of long-term experimental evolution in glutamine-limited chemostats.

https://cdn.elifesciences.org/articles/98934/elife-98934-fig2-data2-v1.pdf

Download elife-98934-fig2-data2-v1.pdf

Decreased CNV formation rates in modified genomes suggests adjacent genomic elements contribute to GAP1 CNV formation

We used nnSBI to infer CNV formation rates and selection coefficients from the evolutionary dynamics observed in glutamine-limited chemostats (Figure 1B). Previously, nnSBI estimations have been experimentally validated demonstrating its accuracy and reliability (Avecilla et al., 2022). First, we trained a neural density estimator using evolutionary simulations (Methods). This neural density estimator then allows us to infer posterior distributions and estimate the model parameters (i.e. the GAP1 CNV formation rate, ${\delta }_C$; the GAP1 CNV selection coefficient, $s_C$) from a single population CNV dynamics. We also inferred a collective posterior distribution from a set of replicate populations of the same strain. This collective posterior distribution consolidates estimations of the formation rate and selection coefficient from multiple replicate populations of one strain into one single estimate per strain in order to compare between the four strains, rather than between all 27 populations. We evaluated the confidence of our inference approach on synthetic simulations by computing its coverage, that is the probability that the true parameter falls within the 95% highest density interval (HDI) of the posterior distribution, a measure of certainty in an estimate similar to confidence intervals (Kruschke, 2021; Supplementary file 2). We find that the posterior distributions are narrow as the 95% HDI are less than an order of magnitude for both $s_C$ and ${\delta }_C$. Thus, we did not apply post-training adjustments to the neural density estimator, such as calibration (Cook et al., 2006) or ensembles (Caspi et al., 2023; Hermans et al., 2022) when estimating ${\delta }_C$ and $s_C$ from experimental GAP1 CNV dynamics.

We find that the individual maximum a posteriori (MAP) estimates vary across strains and replicates (Figure 3—figure supplement 2). Overall, the CNV selection coefficient, ${\displaystyle s_C}$ , ranges from 0.1 to 0.22 (with one exception of 0.3), whereas the CNV formation rate, ${\displaystyle {\delta }_C}$, ranges from ${10}^{-6}$ to ${10}^{-4}$ (with one exception of ${10}^{-3}$ and two of ${10}^{-7}$); and the proportion of early-occurring GAP1 CNVs without amplification of the reporter ($\phi )$ ranges from ${10}^{-6}$ to ${10}^{-2}$ (with two exceptions of ${10}^{-8}$). We found that MAP estimates of replicate populations of the same strain cluster together, with some outliers (Figure 3—figure supplement 2). We performed posterior predictive checks, drawing parameter values from the posterior distributions and simulating the CNV dynamics (Figure 3—figure supplement 3), which agree with the observed data (Figure 3—figure supplement 1). For each strain, we use all individual posterior distributions to infer the collective posterior distribution, which is a posterior distribution conditioned on all observations, $P\left(\theta \right|X_1,\dots ,X_n)$ (Methods). The collective posterior allows us to estimate whether there is a difference in CNV formation rate and fitness effect across the four strains.

Collective posterior HDIs are very narrow (Figure 3A), and samples are highly correlated, as expected for joint estimation of selection coefficients and beneficial mutation rates (Gitschlag et al., 2023). The collective MAP estimates of the CNV selection coefficient are similar for the WT and LTRΔ (0.182). For ARSΔ and ALLΔ, the selection coefficient is estimated to be lower, with values of 0.146 and 0.126, respectively. However, all four selection coefficients are still large, consistent with these populations containing GAP1 CNVs that are highly beneficial under glutamine-limitation. The collective MAP estimate for the CNV formation rate in WT is $4.5\cdot {10}^{-5}$. By contrast, the CNV formation rate is markedly lower in all mutant strains ranging from ${1\cdot 10}^{-5}$ for LTRΔ and ALLΔ to $2.4\cdot {10}^{-6}$ in ARSΔ. These results support our hypothesis that proximate sequence features facilitate GAP1 CNV formation.

Figure 3 with 8 supplements see all

Download asset Open asset

The collective MAP predictions reproduce the experimental observations. Other than the very final time point for the WT population, all collective MAP predictions lay within the interquartile ranges (Figure 3—figure supplement 1). The observed GAP1 CNV proportion stabilizes at different levels in the different experiments (Figure 1B). This can be explained by pre-existing or early-occurring unreported CNVs with proportions estimated to be between $\phi$=$4\cdot {10}^{-6}$ to ${1.6\cdot 10}^{-4}$ by the collective MAPs (Figure 3—figure supplement 4). Indeed, our model predicts that the total (reported and unreported) final CNV proportion is nearly one in all cases (Figure 3—figure supplement 5).

We sought to understand the consequences of differences in CNV formation rate and CNV fitness effects between the four strains on evolutionary dynamics. We used a modified version of the evolutionary model with the estimated parameters to simulate an evolutionary competition between WT and the three architecture mutant strains over 116 generations, a point at which CNVs have reached high proportions in the experiment. To ‘win’ these competitions, the competitor strains need to adapt to glutamine-limitation by producing CNVs. The results of the simulated competitions predict that the WT outcompetes the other strains in all cases as its predicted final proportion almost always exceeds its initial proportion of 0.5 (Figure 3—figure supplement 6 and Supplementary file 3). The average predicted proportion of WT cells when competing with LTRΔ is 0.717. By contrast, ARSΔ and ALLΔ are predicted to be almost eliminated by generation 116, as the average predicted WT proportion is 0.998 and 0.999, respectively. These simulated competitions further suggest that the ARS is a more important contributor to adaptive evolution mediated by GAP1 CNVs.

Next, we estimated de novo CNV diversity in each strain. Previous work showed a diversity of CNV alleles formed under glutamine-limited selection including tandem duplications, segmental amplification, translocations, and whole chromosome amplification (Lauer et al., 2018), and that lineage richness decreases rapidly over the course of evolution due in part to competition and clonal interference (Lauer et al., 2018; Levy et al., 2015; Nguyen Ba et al., 2019). Our model does not include competition, clonal interference, or recurrent CNV formation. Therefore, diversity calculations are likely overestimations. Nonetheless, a comparison of diversity between strains is informative of whether proximate genome elements affect CNV allele diversity. Therefore, for each strain, we used its collective MAP to simulate a posterior prediction for the genotype frequencies (Figure 3C), which we then used to predict the posterior Shannon diversity (Jost, 2006). In all populations, we predict the set of CNV alleles to be highly diverse: the final predicted Shannon diversity ranges from $1.6\cdot {10}^4$ in ARSΔ to $3.2\cdot {10}^5$ in WT (Figure 3B). Our model predicts that the diversity increases rapidly during the selection phase and stabilizes in the equilibrium phase. This is because CNV alleles that form towards the end of the experiment would have a low frequency with a minor effect on diversity. We observe the greatest diversity in WT populations with lower diversity in the three genomic architecture mutants. Moreover, diversity saturates faster in WT populations. This suggests that the WT strain is able to form more unique CNVs allele types earlier compared to the other three strains (Figure 3B). Shannon diversity is lower in LTRΔ and further lower in ALLΔ and ARSΔ (Figure 3B) reflecting the rank order of CNV formation rates (Figure 3A).

Inference of CNV mechanisms in genome architecture mutants

Contrary to our expectations, removal of proximate genomic elements from the GAP1 locus does not inhibit the formation of GAP1 CNVs. We sought to determine the molecular basis by which GAP1 CNVs form in the absence of these local elements. Therefore, we isolated 177 GAP1 CNV-containing clones across each population containing the four different strains at generations 79 and 125 and performed Illumina whole-genome sequencing. Using a combination of read depth, split read, and discordant read analysis, we defined the extent of the amplified region, the precise CNV breakpoints, and GAP1 copy number. On the basis of these features, we inferred the CNV-forming mechanisms for each GAP1 CNV (Methods). Among the 177 analyzed GAP1 CNVs, we observed tandem amplifications, tandem triplications with an inverted middle copy, intra- and inter-chromosomal translocations, aneuploidy, and complex CNVs. GAP1 copy numbers range from two to six in any given clone. Each of the four strains is able to produce a diversity of CNV alleles ranging from small (tens of kilobases) to large (~hundreds of kilobases) segmental amplifications (Figure 4). We quantified the CNV length per strain (Figure 4E) and found no significant relationship between CNV length and generation from which the clone was isolated (ANOVA, p=0.33) and therefore considered all 177 clones in subsequent comparisons (Figure 4—figure supplement 1). We found no significant effect of the inferred $s_C$ (ANOVA, p=0.673) or ${\delta }_C$ (ANOVA, p=0.277) on CNV length. We defined six major CNV-forming mechanisms across the four strains: ODIRA, LTR NAHR, NAHR, transposon-mediated, complex CNVs, and whole chromosome duplication (aneuploidy) and assigned each CNV allele to one mechanism using diagnostic features of each CNV (Figure 4A–D and Methods).

Figure 4 with 3 supplements see all

Download asset Open asset

ODIRA is a predominant mechanism of CNV formation

We inferred GAP1 CNVs formed through ODIRA in all four genotypes at high frequencies: 22 out of 37 WT clones (59%), 42 out of 52 LTR∆ clones (81%), 11 out of 42 ARS∆ clones (26%), and 12 out of 46 ALL∆ clones (26%). Considering the set of all CNVs in all strains, ODIRA is the most common CNV mechanism comprising almost half of all CNVs (87/177, 49%). The second most common mechanism occurs about half as oftenーNAHR between flanking LTRs (38/177, 21%), which generates tandem amplifications. In the WT background, ODIRA (22/37) and NAHR between LTRs (11/37) account for 89% of GAP1 CNVs.

In LTR∆ populations, GAP1 CNVs form via ODIRA, chromosome missegregation, and NAHR using other sites. As expected, in LTR∆ clones we did not detect NAHR between LTRs in 52 clones and no focal amplifications were detected (Figure 5B). In LTR∆ populations CNVs are formed predominantly by ODIRA (42/52, 81%) (Supplementary file 4), a significant increase relative to WT clones (chi-sq, p=0.02469; Figure 4F). By contrast, aneuploidy (5/52), complex CNV (3/52), and NAHR (2/52) account for less than 10% of GAP1 CNVs in LTR∆. Consequently, we observe an increase in average GAP1 CNV length in LTR∆ relative to WT (Figure 4E) as there is an increased prevalence of segmental amplifications and aneuploidy (Figure 5B).

Figure 5

Download asset Open asset

Overall, aneuploidy was observed infrequently. Whole amplification of chromosome XI was detected in six out of 177 clones (3.4%; Figure 4F) and detected in only two strains: WT and LTR∆ (Figure 5A). We also detected supernumerary chromosomes in five out of 177 clones (2.8%), which formed through both NAHR and ODIRA (Figure 5A).

In this study, we sought to understand the molecular basis of repeated de novo amplifications and selection of the general amino acid permease gene, GAP1, in S. cerevisiae evolving under glutamine-limited selection. We hypothesized that a high formation rate of GAP1 CNVs is due to the unique genomic architecture at the locus, which comprises two flanking long terminal repeats and a DNA replication origin. We used genetic engineering, experimental evolution, and neural network simulation-based inference to quantify de novo CNV dynamics and estimate the CNV formation rate and selection coefficient in engineered mutants lacking the proximate genome elements. We find that removal of these elements has a significant impact on de novo CNV dynamics, CNV formation rate, and selection coefficients. However, CNVs are formed and selected in the absence of these elements highlighting the plasticity of the genome and diversity of mechanisms that generate CNVs during adaptive evolution.

Despite their proximity to GAP1 and previous studies demonstrating the prevalence of NAHR between repetitive sequences forming CNVs (Dunham et al., 2002; Gresham et al., 2010; Todd et al., 2019; Zhang et al., 2013), we found that flanking LTRs are not an essential driver of CNV formation. The de novo CNV dynamics of WT and LTR∆ populations are similar and we find that although the CNV formation rate is reduced, the effect is small. By contrast, a significantly decreased CNV formation rate and delayed CNV appearance time was observed in the absence of the ARS in ARS∆ and ALL∆ populations, which suggests that the local ARS is an important determinant of GAP1 CNV-mediated adaptive dynamics. Furthermore, the significant delay in the time at which the CNV frequency reaches equilibrium in the ARS∆ compared to WT (Figure 1E) can be explained by both the estimated lower CNV formation rate and lower selection coefficient (Figure 3A). ODIRA was identified as the predominant CNV mechanism in sequence-characterized clones revealing that DNA replication errors, specifically template switching of the leading and lagging strands, are a common source of CNV formation during adaptive evolution.

Using nnSBI we inferred lower rates of CNV formation in all strains with modified genomes that may be informative of the rate at which specific mechanisms occur. The lower CNV formation rate in the LTR∆ strain could be a closer approximation of ODIRA formation rates at this locus as ODIRA CNVs are the predominant CNV mechanism in the LTR∆ strain (Figure 4F). Furthermore, the low formation rates in the LTR∆ relative to WT might suggest that the presence of the flanking long terminal repeats may increase the rate of ODIRA formation through an otherwise unknown combinatorial effect of DNA replication across these flanking LTRs and template switching at the GAP1 locus. ARS∆ has the lowest CNV formation rate and it could be an approximation of the rates of NAHR between flanking LTRs and ODIRA at distal origins. We find that the ALL∆ has a higher CNV formation rate than the ARS∆. One explanation for this is that the deletion of the flanking LTRs in ALL∆ gives opportunity for novel transposon insertions and subsequent CNV formation through LTR NAHR. Indeed, we find an enrichment of novel transposon-insertions in the ALL∆ (Figure 4F) and subsequent CNV formation through recombination of the Ty1-associated repeats (Figure 5B, ALL∆). The sequential events of transposon insertion followed by LTR NAHR must occur at a high rate to explain the increased CNV rate in ALL∆ relative to ARS∆. While remarkable, increased transposon activity is associated with nutrient stress (Curcio and Garfinkel, 1999; Lesage and Todeschini, 2005; Todeschini et al., 2005) and therefore this is a plausible explanation for the CNV rate estimated in ALL∆. Additionally, ARS∆ clones rely more on LTR NAHR to form CNVs (Figure 4F). The prevalence of ODIRA in ARS∆ and ALL∆ are similar. LTR NAHR usually occurs after double strand breaks at the long terminal repeats to give rise to CNVs (Argueso et al., 2008). Because we use haploid cells, such double strand break and homology-mediated repair would have to occur during S-phase after DNA replication with a sister chromatid repair template to form tandem duplications. Therefore the dependency on LTR NAHR to form CNVs and the spatial (breaks at LTR sequences) and temporal (S-phase) constraints could explain the lower formation rate in ARS∆.

The genomic elements have clear effects on the evolutionary dynamics using simulated competitive fitness experiments. The similar selection coefficients in WT and LTR∆ suggest that CNV clones formed in these background strains are similar. Indeed, the predominant CNV mechanism in both is ODIRA followed by LTR NAHR (Figure 4F). Whereas LTR NAHR is abolished in the LTR∆, it seems that CNVs formed by ODIRA allow adaptation to glutamine-limitation similar to WT. The lower selection coefficients in ARS∆ and ALL∆ suggest that GAP1 CNVs formed in these strains have some cost. In a competition, they would get outcompeted by CNV alleles in the WT and LTR∆ background (Figure 3—figure supplement 6). Additionally, the local ARS, ARS1116, is a major origin (McGuffee et al., 2013) and ODIRA CNVs found around this origin corroborate its activity. The simulated competitions (Figure 3—figure supplement 6) further suggest that the ARS is a more important contributor to adaptive evolution mediated by GAP1 CNVs.

The prevalence of ODIRA generated CNVs is a consequence of multiple DNA replication origins and pervasive inverted repeat sequences throughout the chromosome (Figure 4). In particular, breakpoint analysis of LTR∆ CNV clones show that ODIRA produces a continuum of CNV sizes along the short arm of chromosome XI. Downstream breakpoints of ODIRA-generated CNVs range from nearby the GAP1 gene (~3 kilobases) to the right telomere of chromosome XI (153 kilobases; Figure 5B). The S. cerevisiae genome contains a high frequency of inverted repeats ranging from 3bp to 14bp throughout the genome (Martin et al., 2024), but longer repeats are more likely to be used in ODIRA (Martin et al., 2024). The ubiquity of inverted repeats is in stark contrast to the relative paucity of LTR sequences, which are dispersed throughout the genome. Thus, ODIRA supplies a diverse and high number of gene amplifications for selection to act on, setting the stage for genome evolution and adaptation. It appears that complex CNVs may include secondary rearrangements after an initial ODIRA event as most complex CNVs in this study include signatures of ODIRA events (Supplementary file 6). Further work needs to be done to resolve the CNV structure.

Consistent with previous reports of increased Ty insertions in S. cerevisiae under stress conditions (Morillon et al., 2000; Morillon et al., 2002), we observed novel retrotransposon insertions in populations evolved in glutamine-limited chemostats. Transposon insertions can be harmful and lead to loss-of-function mutations but are also a means of generating beneficial alleles including CNVs (Blanc and Adams, 2003; Dunham et al., 2002; Gresham et al., 2008; Wilke and Adams, 1992). We only detected novel Ty insertions in the ALL∆ strain. This is likely because regions upstream of tRNA genes are predisposed to transposition. Our detection of novel retrotransposon insertions is consistent with a previous experimental evolution study that suggested that Ty insertions were rare under constant nitrogen-limitation and substantially more common under fluctuating nitrogen limitation, in which cells experience total nitrogen starvation periodically (Hays et al., 2023). In that study, 898 novel Ty insertions were found across 345 clones (Hays et al., 2023) corresponding to an average of 2.6 insertions per genome This high insertion frequency is consistent with detecting novel insertions on either side of the GAP1 gene that subsequently mediate a focal amplification via LTR NAHR. Importantly, the role of Ty differs in the two studies, as in our case beneficial CNV formed after novel retrotransposition through recombination of newly introduced repeat sequences, whereas Hays et al., 2023 found Ty-associated null alleles that are beneficial in nitrogen-limited conditions. Together, these results reveal the different means by which retrotransposition can facilitate adaptive evolution.

Aneuploidy was not a major source of adaptation in our experiments as it was infrequently detected (n=6/177). This contrasts with studies suggesting aneuploidy is a rapid and transient route to adaptation over short evolutionary time scales (Chen et al., 2012a; Chen et al., 2015; Chen et al., 2012b; Pavelka et al., 2010; Selmecki et al., 2006; Selmecki et al., 2015; Yona et al., 2012). However, aneuploidy incurs a fitness cost (Robinson et al., 2023; Tsai et al., 2019; Yang et al., 2021) and therefore can be outcompeted by slow-forming but less costly beneficial mutations in large populations (Kohanovski et al., 2024). Our observed higher frequencies of focal and segmental amplifications may be because they are less costly than whole-chromosome amplifications.

A variety of DNA replication errors generate CNVs. Replication slippage at palindromic DNA and DNA repeats can cause fork stalling and downstream CNV formation (Lee et al., 2007; Zhang et al., 2009b). DNA repeats can form secondary structures like R loops, cruciforms, non-B DNA structures, and hairpins which stimulate CNV formation (Gu et al., 2008). Untimely replication, faulty fork progression, S-phase checkpoint dysfunction, defective nucleosome assembly, and DNA repeat sites including LTRs are sources of replication-associated genome instability (Aguilera and García-Muse, 2013).

Additional processes may also play a role. The GAP1 gene is highly transcribed under glutamine-limitation (Airoldi et al., 2016) and transcription-replication collisions may fuel ODIRA CNV formation at this locus (Lauer and Gresham, 2019; Wilson et al., 2015). CNV formation can also be stimulated by transcription-associated replication stress and histone acetylation (Hull et al., 2017; Salim et al., 2021; Whale et al., 2022) and replication fork stalling at tRNA genes (Osmundson et al., 2017; Yeung and Smith, 2020). Testing the role of transcription in promoting the formation of adaptive CNVs warrants further investigation.

Recent work has proposed that ODIRA CNVs are a major mechanism of CNVs in human genomes (Brewer et al., 2011; Brewer et al., 2015; Martin et al., 2024). Studies of human and yeast genomes have typically considered homologous recombination as the predominant mechanism of CNV formation (Lupski and Stankiewicz, 2005). CNV hotspots identified in the human (Chance et al., 1994; Lupski, 1998; Lupski and Stankiewicz, 2005; Pentao et al., 1992) and yeast genomes are indeed mediated by NAHR of long repeat sequences (Green et al., 2010; Gresham et al., 2010). However, a focus on recombination-based mechanisms as a means of generating copy number variation may be the result of ascertainment bias or the comparative ease of studying the effect of long repeat sequences over short palindromic ones. Our study demonstrates that experimental evolution in yeast is a useful approach to elucidating the molecular mechanisms by which DNA replication errors generate CNVs.

Sequencing data is available at SRA PRJNA1098800.Other associated data are available here: https://osf.io/js7z8/. Source code repository simulation-based inference: https://github.com/yoavram-lab/chuong_et_al (copy archived at Ben Nun, 2024). Scripts for flow cytometry-based evolutionary dynamics and analysis of CNV clones: https://github.com/GreshamLab/local_arch_variants (copy archived at Chuong, 2024). Whole genome, split, discordants read depth profiles in the form of .bam files for each CNV strain and their corresponding ancestor aligned to our custom GFP GAP1 reference strain are displayed on https://jbrowse.bio.nyu.edu/gresham/?data=data/ee_gap1_arch_muts for WT strains, https://jbrowse.bio.nyu.edu/gresham/LTRKO_clones for LTRΔ strains, https://jbrowse.bio.nyu.edu/gresham/ARSKO_clones for ARSΔ strains, https://jbrowse.bio.nyu.edu/gresham/ALLKO_clones for ALLΔ strains.CNV breakpoints and associated information for all 177 clones are available in Supplementary file 2.

1. 1. Chuong J
   2. Gresham D

   (2024) NCBI BioProject

   ID PRJNA1098800. Local genome architecture on copy number variant dynamics.

   https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA1098800
2. 1. Chuong J

   (2024) Open Science Framework

   ID js7z8. DNA replication errors are a major source of adaptive gene amplification.

   https://osf.io/js7z8/

1. 1. Aguilera A
   2. García-Muse T

   (2013) Causes of genome instability

   Annual Review of Genetics 47:1–32.

   https://doi.org/10.1146/annurev-genet-111212-133232

   • PubMed
   • Google Scholar
2. 1. Airoldi EM
   2. Miller D
   3. Athanasiadou R
   4. Brandt N
   5. Abdul-Rahman F
   6. Neymotin B
   7. Hashimoto T
   8. Bahmani T
   9. Gresham D

   (2016) Steady-state and dynamic gene expression programs in Saccharomyces cerevisiae in response to variation in environmental nitrogen

   Molecular Biology of the Cell 27:1383–1396.

   https://doi.org/10.1091/mbc.E14-05-1013

   • PubMed
   • Google Scholar
3. 1. Argueso JL
   2. Westmoreland J
   3. Mieczkowski PA
   4. Gawel M
   5. Petes TD
   6. Resnick MA

   (2008) Double-strand breaks associated with repetitive DNA can reshape the genome

   PNAS 105:11845–11850.

   https://doi.org/10.1073/pnas.0804529105

   • PubMed
   • Google Scholar
4. 1. Arlt MF
   2. Wilson TE
   3. Glover TW

   (2012) Replication stress and mechanisms of CNV formation

   Current Opinion in Genetics & Development 22:204–210.

   https://doi.org/10.1016/j.gde.2012.01.009

   • Google Scholar
5. 1. Arndt PF
   2. Hwa T
   3. Petrov DA

   (2005) Substantial regional variation in substitution rates in the human genome: importance of GC content, gene density, and telomere-specific effects

   Journal of Molecular Evolution 60:748–763.

   https://doi.org/10.1007/s00239-004-0222-5

   • PubMed
   • Google Scholar
6. 1. Avecilla G
   2. Chuong JN
   3. Li F
   4. Sherlock G
   5. Gresham D
   6. Ram Y

   (2022) Neural networks enable efficient and accurate simulation-based inference of evolutionary parameters from adaptation dynamics

   PLOS Biology 20:e3001633.

   https://doi.org/10.1371/journal.pbio.3001633

   • PubMed
   • Google Scholar
7. 1. Avecilla G
   2. Spealman P
   3. Matthews J
   4. Caudal E
   5. Schacherer J
   6. Gresham D

   (2023) Copy number variation alters local and global mutational tolerance

   Genome Research 33:1340–1353.

   https://doi.org/10.1101/gr.277625.122

   • PubMed
   • Google Scholar
8. 1. Baym M
   2. Kryazhimskiy S
   3. Lieberman TD
   4. Chung H
   5. Desai MM
   6. Kishony R

   (2015) Inexpensive multiplexed library preparation for megabase-sized genomes

   PLOS ONE 10:e0128036.

   https://doi.org/10.1371/journal.pone.0128036

   • PubMed
   • Google Scholar
9. 1. Ben-David U
   2. Amon A

   (2020) Context is everything: aneuploidy in cancer

   Nature Reviews. Genetics 21:44–62.

   https://doi.org/10.1038/s41576-019-0171-x

   • PubMed
   • Google Scholar
10. Software

    1. Ben Nun N

    (2024) Chuong_et_al, version swh:1:rev:9e7dcecbd3c6334fe0595e554312971c68e89484

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:e4dde7ea04c7f3df0aa84e11e860d0cec76bc5d5;origin=https://github.com/yoavram-lab/chuong_et_al;visit=swh:1:snp:b1bb45357a60405c4a3cff2f854478e092906dd2;anchor=swh:1:rev:9e7dcecbd3c6334fe0595e554312971c68e89484
11. 1. Blanc VM
    2. Adams J

    (2003) Evolution in Saccharomyces cerevisiae: identification of mutations increasing fitness in laboratory populations

    Genetics 165:975–983.

    https://doi.org/10.1093/genetics/165.3.975

    • PubMed
    • Google Scholar
12. 1. Blount ZD
    2. Borland CZ
    3. Lenski RE

    (2008) Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli

    PNAS 105:7899–7906.

    https://doi.org/10.1073/pnas.0803151105

    • Google Scholar
13. 1. Blount ZD
    2. Barrick JE
    3. Davidson CJ
    4. Lenski RE

    (2012) Genomic analysis of a key innovation in an experimental Escherichia coli population

    Nature 489:513–518.

    https://doi.org/10.1038/nature11514

    • PubMed
    • Google Scholar
14. 1. Brewer BJ
    2. Payen C
    3. Raghuraman MK
    4. Dunham MJ

    (2011) Origin-dependent inverted-repeat amplification: a replication-based model for generating palindromic amplicons

    PLOS Genetics 7:e1002016.

    https://doi.org/10.1371/journal.pgen.1002016

    • PubMed
    • Google Scholar
15. 1. Brewer BJ
    2. Payen C
    3. Di Rienzi SC
    4. Higgins MM
    5. Ong G
    6. Dunham MJ
    7. Raghuraman MK

    (2015) Origin-dependent inverted-repeat amplification: tests of a model for inverted DNA amplification

    PLOS Genetics 11:e1005699.

    https://doi.org/10.1371/journal.pgen.1005699

    • Google Scholar
16. 1. Brewer BJ
    2. Dunham MJ
    3. Raghuraman MK

    (2024) A unifying model that explains the origins of human inverted copy number variants

    PLOS Genetics 20:e1011091.

    https://doi.org/10.1371/journal.pgen.1011091

    • Google Scholar
17. 1. Cardoso AR
    2. Oliveira M
    3. Amorim A
    4. Azevedo L

    (2016) Major influence of repetitive elements on disease-associated copy number variants (CNVs)

    Human Genomics 10:30.

    https://doi.org/10.1186/s40246-016-0088-9

    • PubMed
    • Google Scholar
18. 1. Carvalho CMB
    2. Pehlivan D
    3. Ramocki MB
    4. Fang P
    5. Alleva B
    6. Franco LM
    7. Belmont JW
    8. Hastings PJ
    9. Lupski JR

    (2013) Replicative mechanisms for CNV formation are error prone

    Nature Genetics 45:1319–1326.

    https://doi.org/10.1038/ng.2768

    • PubMed
    • Google Scholar
19. 1. Caspi I
    2. Meir M
    3. Ben Nun N
    4. Abu Rass R
    5. Yakhini U
    6. Stern A
    7. Ram Y

    (2023) Mutation rate, selection, and epistasis inferred from RNA virus haplotypes via neural posterior estimation

    Virus Evolution 9:vead033.

    https://doi.org/10.1093/ve/vead033

    • PubMed
    • Google Scholar
20. 1. Chance PF
    2. Abbas N
    3. Lensch MW
    4. Pentao L
    5. Roa BB
    6. Patel PI
    7. Lupski JR

    (1994) Two autosomal dominant neuropathies result from reciprocal DNA duplication/deletion of a region on chromosome 17

    Human Molecular Genetics 3:223–228.

    https://doi.org/10.1093/hmg/3.2.223

    • PubMed
    • Google Scholar
21. 1. Chen G
    2. Bradford WD
    3. Seidel CW
    4. Li R

    (2012a) Hsp90 stress potentiates rapid cellular adaptation through induction of aneuploidy

    Nature 482:246–250.

    https://doi.org/10.1038/nature10795

    • PubMed
    • Google Scholar
22. 1. Chen G
    2. Rubinstein B
    3. Li R

    (2012b) Whole chromosome aneuploidy: big mutations drive adaptation by phenotypic leap

    BioEssays 34:893–900.

    https://doi.org/10.1002/bies.201200069

    • Google Scholar
23. 1. Chen G
    2. Mulla WA
    3. Kucharavy A
    4. Tsai HJ
    5. Rubinstein B
    6. Conkright J
    7. McCroskey S
    8. Bradford WD
    9. Weems L
    10. Haug JS
    11. Seidel CW
    12. Berman J
    13. Li R

    (2015) Targeting the adaptability of heterogeneous aneuploids

    Cell 160:771–784.

    https://doi.org/10.1016/j.cell.2015.01.026

    • Google Scholar
24. 1. Cheng Z
    2. Ventura M
    3. She X
    4. Khaitovich P
    5. Graves T
    6. Osoegawa K
    7. Church D
    8. DeJong P
    9. Wilson RK
    10. Pääbo S
    11. Rocchi M
    12. Eichler EE

    (2005) A genome-wide comparison of recent chimpanzee and human segmental duplications

    Nature 437:88–93.

    https://doi.org/10.1038/nature04000

    • Google Scholar
25. 1. Chuang JH
    2. Li H

    (2004) Functional bias and spatial organization of genes in mutational hot and cold regions in the human genome

    PLOS Biology 2:E29.

    https://doi.org/10.1371/journal.pbio.0020029

    • PubMed
    • Google Scholar
26. Software

    1. Chuong JN

    (2024) Local_arch_variants, version swh:1:rev:804f04ab1b6ed647bedac4a0552df06aee9c22c5

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:a4fc195561c29716cbfd0ccd18b7022cdedcc4f0;origin=https://github.com/GreshamLab/local_arch_variants;visit=swh:1:snp:b0f457af331dabcee18700fb1279db8ffd2be8a4;anchor=swh:1:rev:804f04ab1b6ed647bedac4a0552df06aee9c22c5
27. 1. Cook SR
    2. Gelman A
    3. Rubin DB

    (2006) Validation of software for bayesian models using posterior quantiles

    Journal of Computational and Graphical Statistics 15:675–692.

    https://doi.org/10.1198/106186006X136976

    • Google Scholar
28. 1. Cranmer K
    2. Brehmer J
    3. Louppe G

    (2020) The frontier of simulation-based inference

    PNAS 117:30055–30062.

    https://doi.org/10.1073/pnas.1912789117

    • PubMed
    • Google Scholar
29. 1. Curcio MJ
    2. Garfinkel DJ

    (1999) New lines of host defense: inhibition of Ty1 retrotransposition by Fus3p and NER/TFIIH

    Trends in Genetics 15:43–45.

    https://doi.org/10.1016/s0168-9525(98)01643-6

    • PubMed
    • Google Scholar
30. 1. Di Rienzi SC
    2. Collingwood D
    3. Raghuraman MK
    4. Brewer BJ

    (2009) Fragile genomic sites are associated with origins of replication

    Genome Biology and Evolution 1:350–363.

    https://doi.org/10.1093/gbe/evp034

    • PubMed
    • Google Scholar
31. 1. Dunham MJ
    2. Badrane H
    3. Ferea T
    4. Adams J
    5. Brown PO
    6. Rosenzweig F
    7. Botstein D

    (2002) Characteristic genome rearrangements in experimental evolution of Saccharomyces cerevisiae

    PNAS 99:16144–16149.

    https://doi.org/10.1073/pnas.242624799

    • PubMed
    • Google Scholar
32. 1. Gitschlag BL
    2. Cano AV
    3. Payne JL
    4. McCandlish DM
    5. Stoltzfus A

    (2023) Mutation and Selection Induce Correlations between Selection Coefficients and Mutation Rates

    The American Naturalist 202:534–557.

    https://doi.org/10.1086/726014

    • PubMed
    • Google Scholar
33. 1. Gonçalves PJ
    2. Lueckmann JM
    3. Deistler M
    4. Nonnenmacher M
    5. Öcal K
    6. Bassetto G
    7. Chintaluri C
    8. Podlaski WF
    9. Haddad SA
    10. Vogels TP
    11. Greenberg DS
    12. Macke JH

    (2020) Training deep neural density estimators to identify mechanistic models of neural dynamics

    eLife 9:e56261.

    https://doi.org/10.7554/eLife.56261

    • PubMed
    • Google Scholar
34. 1. Green BM
    2. Finn KJ
    3. Li JJ

    (2010) Loss of DNA replication control is a potent inducer of gene amplification

    Science 329:943–946.

    https://doi.org/10.1126/science.1190966

    • PubMed
    • Google Scholar
35. 1. Gresham D
    2. Desai MM
    3. Tucker CM
    4. Jenq HT
    5. Pai DA
    6. Ward A
    7. DeSevo CG
    8. Botstein D
    9. Dunham MJ

    (2008) The repertoire and dynamics of evolutionary adaptations to controlled nutrient-limited environments in yeast

    PLOS Genetics 4:e1000303.

    https://doi.org/10.1371/journal.pgen.1000303

    • PubMed
    • Google Scholar
36. 1. Gresham D
    2. Usaite R
    3. Germann SM
    4. Lisby M
    5. Botstein D
    6. Regenberg B

    (2010) Adaptation to diverse nitrogen-limited environments by deletion or extrachromosomal element formation of the GAP1 locus

    PNAS 107:18551–18556.

    https://doi.org/10.1073/pnas.1014023107

    • PubMed
    • Google Scholar
37. 1. Gu W
    2. Zhang F
    3. Lupski JR

    (2008) Mechanisms for human genomic rearrangements

    PathoGenetics 1:4.

    https://doi.org/10.1186/1755-8417-1-4

    • PubMed
    • Google Scholar
38. 1. Hall DW
    2. Mahmoudizad R
    3. Hurd AW
    4. Joseph SB

    (2008) Spontaneous mutations in diploid Saccharomyces cerevisiae: another thousand cell generations

    Genetics Research 90:229–241.

    https://doi.org/10.1017/S0016672308009324

    • PubMed
    • Google Scholar
39. Book

    1. Harel T
    2. Pehlivan D
    3. Caskey CT
    4. Lupski JR

    (2015) Chapter 1—mendelian, non-mendelian, multigenic inheritance, and epigenetics

    In: Rosenberg RN, Pascual JM, editors. Rosenberg’s Molecular and Genetic Basis of Neurological and Psychiatric Disease (Fifth Edition). Academic Press. pp. 3–27.

    https://doi.org/10.1016/B978-0-12-410529-4.00001-2

    • Google Scholar
40. 1. Hastings PJ
    2. Ira G
    3. Lupski JR

    (2009a) A microhomology-mediated break-induced replication model for the origin of human copy number variation

    PLOS Genetics 5:e1000327.

    https://doi.org/10.1371/journal.pgen.1000327

    • PubMed
    • Google Scholar
41. 1. Hastings PJ
    2. Lupski JR
    3. Rosenberg SM
    4. Ira G

    (2009b) Mechanisms of change in gene copy number

    Nature Reviews. Genetics 10:551–564.

    https://doi.org/10.1038/nrg2593

    • PubMed
    • Google Scholar
42. 1. Hays M
    2. Schwartz K
    3. Schmidtke DT
    4. Aggeli D
    5. Sherlock G

    (2023) Paths to adaptation under fluctuating nitrogen starvation: the spectrum of adaptive mutations in Saccharomyces cerevisiae is shaped by retrotransposons and microhomology-mediated recombination

    PLOS Genetics 19:e1010747.

    https://doi.org/10.1371/journal.pgen.1010747

    • PubMed
    • Google Scholar
43. Preprint

    1. Hermans J
    2. Delaunoy A
    3. Rozet F
    4. Wehenkel A
    5. Begy V
    6. Louppe G

    (2022) A Trust Crisis In Simulation-Based Inference? Your Posterior Approximations Can Be Unfaithful

    arXiv.

    https://doi.org/10.48550/arXiv.2110.06581

    • Google Scholar
44. 1. Hong J
    2. Gresham D

    (2014) Molecular specificity, convergence and constraint shape adaptive evolution in nutrient-poor environments

    PLOS Genetics 10:e1004041.

    https://doi.org/10.1371/journal.pgen.1004041

    • PubMed
    • Google Scholar
45. 1. Horiuchi T
    2. Horiuchi S
    3. Novick A

    (1963) The genetic basis of hyper-synthesis of beta-galactosidase

    Genetics 48:157–169.

    https://doi.org/10.1093/genetics/48.2.157

    • PubMed
    • Google Scholar
46. 1. Hull RM
    2. Cruz C
    3. Jack CV
    4. Houseley J

    (2017) Environmental change drives accelerated adaptation through stimulated copy number variation

    PLOS Biology 15:e2001333.

    https://doi.org/10.1371/journal.pbio.2001333

    • PubMed
    • Google Scholar
47. 1. Ji H
    2. Moore DP
    3. Blomberg MA
    4. Braiterman LT
    5. Voytas DF
    6. Natsoulis G
    7. Boeke JD

    (1993) Hotspots for unselected Ty1 transposition events on yeast chromosome III are near tRNA genes and LTR sequences

    Cell 73:1007–1018.

    https://doi.org/10.1016/0092-8674(93)90278-x

    • PubMed
    • Google Scholar
48. 1. Joseph SB
    2. Hall DW

    (2004) Spontaneous mutations in diploid Saccharomyces cerevisiae: more beneficial than expected

    Genetics 168:1817–1825.

    https://doi.org/10.1534/genetics.104.033761

    • PubMed
    • Google Scholar
49. 1. Jost L

    (2006) Entropy and diversity

    Oikos 113:363–375.

    https://doi.org/10.1111/j.2006.0030-1299.14714.x

    • Google Scholar
50. 1. Kohanovski I
    2. Pontz M
    3. Vande Zande P
    4. Selmecki A
    5. Dahan O
    6. Pilpel Y
    7. Yona AH
    8. Ram Y

    (2024) Aneuploidy can be an evolutionary diversion on the path to adaptation

    Molecular Biology and Evolution 41:msae052.

    https://doi.org/10.1093/molbev/msae052

    • PubMed
    • Google Scholar
51. 1. Kruschke JK

    (2021) Bayesian analysis reporting guidelines

    Nature Human Behaviour 5:1282–1291.

    https://doi.org/10.1038/s41562-021-01177-7

    • PubMed
    • Google Scholar
52. 1. Lang GI
    2. Botstein D
    3. Desai MM

    (2011) Genetic variation and the fate of beneficial mutations in asexual populations

    Genetics 188:647–661.

    https://doi.org/10.1534/genetics.111.128942

    • PubMed
    • Google Scholar
53. 1. Lang GI
    2. Murray AW

    (2011) Mutation rates across budding yeast chromosome VI are correlated with replication timing

    Genome Biology and Evolution 3:799–811.

    https://doi.org/10.1093/gbe/evr054

    • PubMed
    • Google Scholar
54. 1. Lauer S
    2. Avecilla G
    3. Spealman P
    4. Sethia G
    5. Brandt N
    6. Levy SF
    7. Gresham D

    (2018) Single-cell copy number variant detection reveals the dynamics and diversity of adaptation

    PLOS Biology 16:e3000069.

    https://doi.org/10.1371/journal.pbio.3000069

    • PubMed
    • Google Scholar
55. 1. Lauer S
    2. Gresham D

    (2019) An evolving view of copy number variants

    Current Genetics 65:1287–1295.

    https://doi.org/10.1007/s00294-019-00980-0

    • PubMed
    • Google Scholar
56. 1. Lee JA
    2. Carvalho CMB
    3. Lupski JR

    (2007) A DNA replication mechanism for generating nonrecurrent rearrangements associated with genomic disorders

    Cell 131:1235–1247.

    https://doi.org/10.1016/j.cell.2007.11.037

    • Google Scholar
57. 1. Lercher MJ
    2. Hurst LD

    (2002) Human SNP variability and mutation rate are higher in regions of high recombination

    Trends in Genetics 18:337–340.

    https://doi.org/10.1016/S0168-9525(02)02669-0

    • Google Scholar
58. 1. Lesage P
    2. Todeschini AL

    (2005) Happy together: the life and times of Ty retrotransposons and their hosts

    Cytogenetic and Genome Research 110:70–90.

    https://doi.org/10.1159/000084940

    • PubMed
    • Google Scholar
59. 1. Levy SF
    2. Blundell JR
    3. Venkataram S
    4. Petrov DA
    5. Fisher DS
    6. Sherlock G

    (2015) Quantitative evolutionary dynamics using high-resolution lineage tracking

    Nature 519:181–186.

    https://doi.org/10.1038/nature14279

    • PubMed
    • Google Scholar
60. 1. Lobachev KS
    2. Gordenin DA
    3. Resnick MA

    (2002) The Mre11 complex is required for repair of hairpin-capped double-strand breaks and prevention of chromosome rearrangements

    Cell 108:183–193.

    https://doi.org/10.1016/s0092-8674(02)00614-1

    • PubMed
    • Google Scholar
61. 1. Lukow DA
    2. Sausville EL
    3. Suri P
    4. Chunduri NK
    5. Wieland A
    6. Leu J
    7. Smith JC
    8. Girish V
    9. Kumar AA
    10. Kendall J
    11. Wang Z
    12. Storchova Z
    13. Sheltzer JM

    (2021) Chromosomal instability accelerates the evolution of resistance to anti-cancer therapies

    Developmental Cell 56:2427–2439.

    https://doi.org/10.1016/j.devcel.2021.07.009

    • PubMed
    • Google Scholar
62. 1. Lupski JR

    (1998) Genomic disorders: structural features of the genome can lead to DNA rearrangements and human disease traits

    Trends in Genetics 14:417–422.

    https://doi.org/10.1016/s0168-9525(98)01555-8

    • PubMed
    • Google Scholar
63. 1. Lupski JR
    2. Stankiewicz P

    (2005) Genomic disorders: molecular mechanisms for rearrangements and conveyed phenotypes

    PLOS Genetics 1:e49.

    https://doi.org/10.1371/journal.pgen.0010049

    • PubMed
    • Google Scholar
64. 1. Malhotra D
    2. Sebat J

    (2012) CNVs: harbingers of a rare variant revolution in psychiatric genetics

    Cell 148:1223–1241.

    https://doi.org/10.1016/j.cell.2012.02.039

    • Google Scholar
65. 1. Martin R
    2. Espinoza CY
    3. Large CRL
    4. Rosswork J
    5. Van Bruinisse C
    6. Miller AW
    7. Sanchez JC
    8. Miller M
    9. Paskvan S
    10. Alvino GM
    11. Dunham MJ
    12. Raghuraman MK
    13. Brewer BJ

    (2024) Template switching between the leading and lagging strands at replication forks generates inverted copy number variants through hairpin-capped extrachromosomal DNA

    PLOS Genetics 20:e1010850.

    https://doi.org/10.1371/journal.pgen.1010850

    • PubMed
    • Google Scholar
66. 1. Matassi G
    2. Sharp PM
    3. Gautier C

    (1999) Chromosomal location effects on gene sequence evolution in mammals

    Current Biology 9:786–791.

    https://doi.org/10.1016/s0960-9822(99)80361-3

    • PubMed
    • Google Scholar
67. 1. McGuffee SR
    2. Smith DJ
    3. Whitehouse I

    (2013) Quantitative, genome-wide analysis of eukaryotic replication initiation and termination

    Molecular Cell 50:123–135.

    https://doi.org/10.1016/j.molcel.2013.03.004

    • PubMed
    • Google Scholar
68. 1. Morillon A
    2. Springer M
    3. Lesage P

    (2000) Activation of the Kss1 invasive-filamentous growth pathway induces Ty1 transcription and retrotransposition in Saccharomyces cerevisiae

    Molecular and Cellular Biology 20:5766–5776.

    https://doi.org/10.1128/MCB.20.15.5766-5776.2000

    • PubMed
    • Google Scholar
69. 1. Morillon A
    2. Bénard L
    3. Springer M
    4. Lesage P

    (2002) Differential effects of chromatin and Gcn4 on the 50-fold range of expression among individual yeast Ty1 retrotransposons

    Molecular and Cellular Biology 22:2078–2088.

    https://doi.org/10.1128/MCB.22.7.2078-2088.2002

    • PubMed
    • Google Scholar
70. 1. Mularoni L
    2. Zhou Y
    3. Bowen T
    4. Gangadharan S
    5. Wheelan SJ
    6. Boeke JD

    (2012) Retrotransposon Ty1 integration targets specifically positioned asymmetric nucleosomal DNA segments in tRNA hotspots

    Genome Research 22:693–703.

    https://doi.org/10.1101/gr.129460.111

    • PubMed
    • Google Scholar
71. 1. Narayanan V
    2. Mieczkowski PA
    3. Kim HM
    4. Petes TD
    5. Lobachev KS

    (2006) The pattern of gene amplification is determined by the chromosomal location of hairpin-capped breaks

    Cell 125:1283–1296.

    https://doi.org/10.1016/j.cell.2006.04.042

    • PubMed
    • Google Scholar
72. 1. Nguyen Ba AN
    2. Cvijović I
    3. Rojas Echenique JI
    4. Lawrence KR
    5. Rego-Costa A
    6. Liu X
    7. Levy SF
    8. Desai MM

    (2019) High-resolution lineage tracking reveals travelling wave of adaptation in laboratory yeast

    Nature 575:494–499.

    https://doi.org/10.1038/s41586-019-1749-3

    • Google Scholar
73. 1. Nishant KT
    2. Singh ND
    3. Alani E

    (2009) Genomic mutation rates: What high-throughput methods can tell us

    BioEssays: News and Reviews in Molecular, Cellular and Developmental Biology 31:912–920.

    https://doi.org/10.1002/bies.200900017

    • Google Scholar
74. Book

    1. Ohno S

    (1970)

    Evolution by Gene Duplication

    New York: Springer-Verlag.

    • Google Scholar
75. 1. Osmundson JS
    2. Kumar J
    3. Yeung R
    4. Smith DJ

    (2017) Pif1-family helicases cooperatively suppress widespread replication-fork arrest at tRNA genes

    Nature Structural & Molecular Biology 24:162–170.

    https://doi.org/10.1038/nsmb.3342

    • Google Scholar
76. Preprint

    1. Papamakarios G
    2. Pavlakou T
    3. Murray I

    (2018) Masked Autoregressive Flow for Density Estimation

    arXiv.

    https://doi.org/10.48550/arXiv.1705.07057

    • Google Scholar
77. Preprint

    1. Papamakarios G

    (2019) Neural Density Estimation and Likelihood-Free Inference

    arXiv.

    https://doi.org/10.48550/arXiv.1910.13233

    • Google Scholar
78. 1. Pavelka N
    2. Rancati G
    3. Zhu J
    4. Bradford WD
    5. Saraf A
    6. Florens L
    7. Sanderson BW
    8. Hattem GL
    9. Li R

    (2010) Aneuploidy confers quantitative proteome changes and phenotypic variation in budding yeast

    Nature 468:321–325.

    https://doi.org/10.1038/nature09529

    • PubMed
    • Google Scholar
79. 1. Payen C
    2. Sunshine AB
    3. Ong GT
    4. Pogachar JL
    5. Zhao W
    6. Dunham MJ

    (2016) High-throughput identification of adaptive mutations in experimentally evolved Yeast Populations

    PLOS Genetics 12:e1006339.

    https://doi.org/10.1371/journal.pgen.1006339

    • PubMed
    • Google Scholar
80. 1. Pentao L
    2. Wise CA
    3. Chinault AC
    4. Patel PI
    5. Lupski JR

    (1992) Charcot-Marie-Tooth type 1A duplication appears to arise from recombination at repeat sequences flanking the 1.5 Mb monomer unit

    Nature Genetics 2:292–300.

    https://doi.org/10.1038/ng1292-292

    • PubMed
    • Google Scholar
81. 1. Pös O
    2. Radvanszky J
    3. Buglyó G
    4. Pös Z
    5. Rusnakova D
    6. Nagy B
    7. Szemes T

    (2021) DNA copy number variation: Main characteristics, evolutionary significance, and pathological aspects

    Biomedical Journal 44:548–559.

    https://doi.org/10.1016/j.bj.2021.02.003

    • PubMed
    • Google Scholar
82. 1. Robinson D
    2. Vanacloig-Pedros E
    3. Cai R
    4. Place M
    5. Hose J
    6. Gasch AP

    (2023) Gene-by-environment interactions influence the fitness cost of gene copy-number variation in yeast

    G3\ 13:jkad159.

    https://doi.org/10.1093/g3journal/jkad159

    • PubMed
    • Google Scholar
83. 1. Rutledge SD
    2. Douglas TA
    3. Nicholson JM
    4. Vila-Casadesús M
    5. Kantzler CL
    6. Wangsa D
    7. Barroso-Vilares M
    8. Kale SD
    9. Logarinho E
    10. Cimini D

    (2016) Selective advantage of trisomic human cells cultured in non-standard conditions

    Scientific Reports 6:22828.

    https://doi.org/10.1038/srep22828

    • PubMed
    • Google Scholar
84. 1. Salim D
    2. Bradford WD
    3. Rubinstein B
    4. Gerton JL

    (2021) DNA replication, transcription, and H3K56 acetylation regulate copy number and stability at tandem repeats

    G3 11:jkab082.

    https://doi.org/10.1093/g3journal/jkab082

    • PubMed
    • Google Scholar
85. 1. Selmecki A
    2. Forche A
    3. Berman J

    (2006) Aneuploidy and isochromosome formation in drug-resistant Candida albicans

    Science 313:367–370.

    https://doi.org/10.1126/science.1128242

    • PubMed
    • Google Scholar
86. 1. Selmecki AM
    2. Maruvka YE
    3. Richmond PA
    4. Guillet M
    5. Shoresh N
    6. Sorenson AL
    7. De S
    8. Kishony R
    9. Michor F
    10. Dowell R
    11. Pellman D

    (2015) Polyploidy can drive rapid adaptation in yeast

    Nature 519:349–352.

    https://doi.org/10.1038/nature14187

    • PubMed
    • Google Scholar
87. 1. Sonti RV
    2. Roth JR

    (1989) Role of gene duplications in the adaptation of Salmonella typhimurium to growth on limiting carbon sources

    Genetics 123:19–28.

    https://doi.org/10.1093/genetics/123.1.19

    • PubMed
    • Google Scholar
88. Software

    1. Spealman P

    (2019) CVish, version 1

    Github.

    https://github.com/pspealman/CVish
89. 1. Spealman P
    2. Avecilla G
    3. Matthews J
    4. Suresh I
    5. Gresham D

    (2022) Complex genomic rearrangements following selection in a glutamine-limited medium over hundreds of generations

    Microbiology Resource Announcements 11:e0072922.

    https://doi.org/10.1128/mra.00729-22

    • PubMed
    • Google Scholar
90. 1. Spealman P
    2. De T
    3. Chuong JN
    4. Gresham D

    (2023) Best practices in microbial experimental evolution: using reporters and long-read sequencing to identify copy number variation in experimental evolution

    Journal of Molecular Evolution 91:356–368.

    https://doi.org/10.1007/s00239-023-10102-7

    • PubMed
    • Google Scholar
91. 1. Stankiewicz P
    2. Shaw CJ
    3. Dapper JD
    4. Wakui K
    5. Shaffer LG
    6. Withers M
    7. Elizondo L
    8. Park SS
    9. Lupski JR

    (2003) Genome architecture catalyzes nonrecurrent chromosomal rearrangements

    The American Journal of Human Genetics 72:1101–1116.

    https://doi.org/10.1086/374385

    • Google Scholar
92. 1. Storz JF

    (2016) Gene duplication and evolutionary innovations in hemoglobin-oxygen transport

    Physiology 31:223–232.

    https://doi.org/10.1152/physiol.00060.2015

    • PubMed
    • Google Scholar
93. 1. Taylor JS
    2. Raes J

    (2004) Duplication and divergence: the evolution of new genes and old ideas

    Annual Review of Genetics 38:615–643.

    https://doi.org/10.1146/annurev.genet.38.072902.092831

    • PubMed
    • Google Scholar
94. Preprint

    1. Tejero-Cantero A
    2. Boelts J
    3. Deistler M
    4. Lueckmann JM
    5. Durkan C
    6. Gonçalves PJ
    7. Greenberg DS
    8. Macke JH

    (2020) SBI: A Toolkit for Simulation-Based Inference

    arXiv.

    https://doi.org/10.21105/joss.02505

    • Google Scholar
95. 1. Todd RT
    2. Wikoff TD
    3. Forche A
    4. Selmecki A

    (2019) Genome plasticity in Candida albicans is driven by long repeat sequences

    eLife 8:e45954.

    https://doi.org/10.7554/eLife.45954

    • PubMed
    • Google Scholar
96. 1. Todeschini AL
    2. Morillon A
    3. Springer M
    4. Lesage P

    (2005) Severe adenine starvation activates Ty1 transcription and retrotransposition in Saccharomyces cerevisiae

    Molecular and Cellular Biology 25:7459–7472.

    https://doi.org/10.1128/MCB.25.17.7459-7472.2005

    • PubMed
    • Google Scholar
97. 1. Tsai HJ
    2. Nelliat AR
    3. Choudhury MI
    4. Kucharavy A
    5. Bradford WD
    6. Cook ME
    7. Kim J
    8. Mair DB
    9. Sun SX
    10. Schatz MC
    11. Li R

    (2019) Hypo-osmotic-like stress underlies general cellular defects of aneuploidy

    Nature 570:117–121.

    https://doi.org/10.1038/s41586-019-1187-2

    • PubMed
    • Google Scholar
98. 1. Turner DJ
    2. Miretti M
    3. Rajan D
    4. Fiegler H
    5. Carter NP
    6. Blayney ML
    7. Beck S
    8. Hurles ME

    (2008) Germline rates of de novo meiotic deletions and duplications causing several genomic disorders

    Nature Genetics 40:90–95.

    https://doi.org/10.1038/ng.2007.40

    • Google Scholar
99. 1. Venkataram S
    2. Dunn B
    3. Li Y
    4. Agarwala A
    5. Chang J
    6. Ebel ER
    7. Geiler-Samerotte K
    8. Hérissant L
    9. Blundell JR
    10. Levy SF
    11. Fisher DS
    12. Sherlock G
    13. Petrov DA

    (2016) Development of a comprehensive genotype-to-fitness map of adaptation-driving mutations in yeast

    Cell 166:1585–1596.

    https://doi.org/10.1016/j.cell.2016.08.002

    • PubMed
    • Google Scholar
100. 1. Whale AJ
     2. King M
     3. Hull RM
     4. Krueger F
     5. Houseley J

     (2022) Stimulation of adaptive gene amplification by origin firing under replication fork constraint

     Nucleic Acids Research 50:915–936.

     https://doi.org/10.1093/nar/gkab1257

     • PubMed
     • Google Scholar
101. 1. Wilke CM
     2. Adams J

     (1992) Fitness effects of Ty transposition in Saccharomyces cerevisiae

     Genetics 131:31–42.

     https://doi.org/10.1093/genetics/131.1.31

     • PubMed
     • Google Scholar
102. 1. Wilson TE
     2. Arlt MF
     3. Park SH
     4. Rajendran S
     5. Paulsen M
     6. Ljungman M
     7. Glover TW

     (2015) Large transcription units unify copy number variants and common fragile sites arising under replication stress

     Genome Research 25:189–200.

     https://doi.org/10.1101/gr.177121.114

     • PubMed
     • Google Scholar
103. 1. Wolfe KH
     2. Sharp PM
     3. Li WH

     (1989) Mutation rates differ among regions of the mammalian genome

     Nature 337:283–285.

     https://doi.org/10.1038/337283a0

     • PubMed
     • Google Scholar
104. 1. Yang F
     2. Todd RT
     3. Selmecki A
     4. Jiang Y
     5. Cao Y
     6. Berman J

     (2021) The fitness costs and benefits of trisomy of each Candida albicans chromosome

     Genetics 218:iyab056.

     https://doi.org/10.1093/genetics/iyab056

     • Google Scholar
105. 1. Yeung R
     2. Smith DJ

     (2020) Determinants of Replication-Fork Pausing at tRNA Genes in Saccharomyces cerevisiae

     Genetics 214:825–838.

     https://doi.org/10.1534/genetics.120.303092

     • PubMed
     • Google Scholar
106. 1. Yona AH
     2. Manor YS
     3. Herbst RH
     4. Romano GH
     5. Mitchell A
     6. Kupiec M
     7. Pilpel Y
     8. Dahan O

     (2012) Chromosomal duplication is a transient evolutionary solution to stress

     PNAS 109:21010–21015.

     https://doi.org/10.1073/pnas.1211150109

     • PubMed
     • Google Scholar
107. 1. Zhang F
     2. Gu W
     3. Hurles ME
     4. Lupski JR

     (2009a) Copy number variation in human health, disease, and evolution

     Annual Review of Genomics and Human Genetics 10:451–481.

     https://doi.org/10.1146/annurev.genom.9.081307.164217

     • Google Scholar
108. 1. Zhang F
     2. Khajavi M
     3. Connolly AM
     4. Towne CF
     5. Batish SD
     6. Lupski JR

     (2009b) The DNA replication FoSTeS/MMBIR mechanism can generate genomic, genic and exonic complex rearrangements in humans

     Nature Genetics 41:849–853.

     https://doi.org/10.1038/ng.399

     • Google Scholar
109. 1. Zhang H
     2. Zeidler AFB
     3. Song W
     4. Puccia CM
     5. Malc E
     6. Greenwell PW
     7. Mieczkowski PA
     8. Petes TD
     9. Argueso JL

     (2013) Gene copy-number variation in haploid and diploid strains of the yeast Saccharomyces cerevisiae

     Genetics 193:785–801.

     https://doi.org/10.1534/genetics.112.146522

     • Google Scholar
110. 1. Zuellig MP
     2. Sweigart AL

     (2018) Gene duplicates cause hybrid lethality between sympatric species of Mimulus

     PLOS Genetics 14:e1007130.

     https://doi.org/10.1371/journal.pgen.1007130

     • PubMed
     • Google Scholar

Author details

1. Julie N Chuong

   Department of Biology, Center for Genomics and Systems Biology, New York University, New York, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4388-9458

2. Nadav Ben Nun

   1. School of Zoology, Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
   2. Edmond J. Safra Center for Bioinformatics, Tel Aviv University, Tel Aviv, Israel

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0009-0003-3228-7720

3. Ina Suresh

   Department of Biology, Center for Genomics and Systems Biology, New York University, New York, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

4. Julia Cano Matthews

   Department of Biology, Center for Genomics and Systems Biology, New York University, New York, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

5. Titir De

   Department of Biology, Center for Genomics and Systems Biology, New York University, New York, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

6. Grace Avecilla

   Department of Natural Sciences, Baruch College CUNY, New York, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

7. Farah Abdul-Rahman

   1. Department of Ecology and Evolutionary Biology, Yale University, New Haven, United States
   2. Microbial Sciences Institute, Yale University, New Haven, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

8. Nathan Brandt

   Department of Biological Sciences, North Carolina State University, Raleigh, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

9. Yoav Ram

   1. School of Zoology, Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
   2. Edmond J. Safra Center for Bioinformatics, Tel Aviv University, Tel Aviv, Israel

   Contribution

   Software, Formal analysis, Supervision, Funding acquisition, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9653-4458

10. David Gresham

    Department of Biology, Center for Genomics and Systems Biology, New York University, New York, United States

    Contribution

    Conceptualization, Supervision, Funding acquisition, Investigation, Methodology, Project administration, Writing – review and editing

    For correspondence

    dgresham@nyu.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-4028-0364

• 670

  views

• 17

  downloads

• 0

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.