Search

Predicting gene expression divergence is integral to understanding the emergence of new biological functions and associated traits. Whereas several sophisticated methods have been developed for this task, their applications are either limited to duplicate genes or require expression data from more than two species. Thus, here we present PredIcting eXpression dIvergence (PiXi), the first machine learning framework for predicting gene expression divergence between single-copy orthologs in two species. PiXi models gene expression evolution as an Ornstein-Uhlenbeck process, and overlays this model with multi-layer neural network (NN), random forest, and support vector machine architectures for making predictions. It outputs the predicted class “conserved” or “diverged” for each pair of orthologs, as well as their predicted expression optima in the two species. We show that PiXi has high power and accuracy in predicting gene expression divergence between single-copy orthologs, as well as high accuracy and precision in estimating their expression optima in the two species, across a wide range of evolutionary scenarios, with the globally best performance achieved by a multi-layer NN. Moreover, application of our best-performing PiXi predictor to empirical gene expression data from single-copy orthologs residing at different loci in two species of Drosophila reveals that approximately 23% underwent expression divergence after positional relocation. Further analysis shows that several of these “diverged” genes are involved in the electron transport chain of the mitochondrial membrane, suggesting that new chromatin environments may impact energy production in Drosophila. Thus, by providing a toolkit for predicting gene expression divergence between single-copy orthologs in two species, PiXi can shed light on the origins of novel phenotypes across diverse biological processes and study systems.

Modern comparative biology owes much to phylogenetic regression. At its conception, this technique sparked a revolution that armed biologists with phylogenetic comparative methods (PCMs) for combatting evolutionary pseudoreplication, which arises inherently from trait data sampled across related species. Over the past few decades, the phylogenetic regression framework has become a paradigm of modern comparative biology that has been widely embraced as a remedy for evolutionary pseudoreplication. However, recent evidence has sown doubt over the efficacy of phylogenetic regression, and PCMs more generally, with the suggestion that many of these methods fail to provide an adequate defense against unreplicated evolution—the primary justification for using them in the first place. Importantly, some of the most compelling examples of biological innovation in nature result from abrupt, lineage-specific evolutionary shifts, which current regression models are largely ill-equipped to deal with. Here we explore a solution to this problem by applying robust linear regression to comparative trait data. We formally introduce robust phylogenetic regression to the PCM toolkit with linear estimators that are less sensitive to model violations while still retaining high power to detect true trait associations. Our analyses also highlight an ingenuity of the original algorithm for phylogenetic regression based on independent contrasts, whereby robust estimators are particularly effective. Collectively, we find that robust estimators hold promise for improving tests of trait associations and offer a path forward in scenarios where classical approaches may fail. Our study joins recent arguments for increased vigilance against pseudoreplication and a better understanding of evolutionary model performance in challenging–yet biologically important–settings.

Predicting gene expression divergence and its evolutionary parameters is integral to understanding the emergence of new gene functions and associated traits. Whereas several sophisticated methods have been developed for these tasks, their applications are either limited to duplicate genes or require expression data from more than two species. Thus, here we present PiXi, the first machine learning framework for predicting expression divergence and its evolutionary parameters between single-copy genes in two species. In particular, PiXi models gene expression evolution as an Ornstein-Uhlenbeck process, and overlays this model with multi-layer neural network, random forest, and support vector machine architectures for making predictions. We show that PiXi has high power and accuracy in predicting gene expression divergence and its underlying parameters across a wide range of evolutionary scenarios, with the globally best performance achieved by a multi-layer neural network. Moreover, application of our best performing PiXi predictor to empirical data from single-copy genes residing at different loci in two species of Drosophila reveals that expression divergence occurs in approximately 20% of these positionally relocated genes and is driven by a combination of neutral and selective forces. Further analysis shows that several of these genes are involved in the electron transport chain of the mitochondrial membrane, suggesting that new chromatin environments may impact energy production in Drosophila. Thus, by providing a toolkit for predicting expression divergence and its evolutionary parameters between single-copy genes in two species, PiXi can shed light on the origins of novel phenotypes across diverse biological processes and study systems.

Epistasis is an evolutionary phenomenon whereby the fitness effect of a mutation depends on the genetic background in which it arises. A key source of epistasis in an RNA molecule is its secondary structure, which contains functionally important topological motifs held together by hydrogen bonds between Watson–Crick (WC) base pairs. Here we study epistasis in the secondary structure of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) by examining properties of derived alleles arising from substitution mutations at ancestral WC base-paired and unpaired (UP) sites in 15 conserved topological motifs across the genome. We uncover fewer derived alleles and lower derived allele frequencies at WC than at UP sites, supporting the hypothesis that modifications to the secondary structure are often deleterious. At WC sites, we also find lower derived allele frequencies for mutations that abolish base pairing than for those that yield G·U “wobbles,” illustrating that weak base pairing can partially preserve the integrity of the secondary structure. Last, we show that WC sites under the strongest epistatic constraint reside in a three-stemmed pseudoknot motif that plays an essential role in programmed ribosomal frameshifting, whereas those under the weakest epistatic constraint are located in 3’ UTR motifs that regulate viral replication and pathogenicity. Our findings demonstrate the importance of epistasis in the evolution of the SARS-CoV-2 secondary structure, as well as highlight putative structural and functional targets of different forms of natural selection.

Much of the enormous phenotypic variation observed across human populations is thought to have arisen from events experienced as our ancestors peopled different regions of the world. However, little is known about the genes involved in these population-specific adaptations. Here, we explore this problem by simultaneously examining population-specific genetic and expression differentiation in four human populations. In particular, we derive a branch-based estimator of population-specific differentiation in four populations, and apply this statistic to single-nucleotide polymorphism and RNA-seq data from Italian, British, Finish, and Yoruban populations. As expected, genome-wide estimates of genetic and expression differentiation each independently recapitulate the known relationships among these four human populations, highlighting the utility of our statistic for identifying putative targets of population-specific adaptations. Moreover, genes with large copy number variations display elevated levels of population-specific genetic and expression differentiation, consistent with the hypothesis that gene duplication and deletion events are key reservoirs of adaptive variation. Further, many top-scoring genes are well-known targets of adaptation in Europeans, including those involved in lactase persistence and vitamin D absorption, and a handful of novel candidates represent promising avenues for future research. Together, these analyses reveal that our statistic can aid in uncovering genes involved in population-specific genetic and expression differentiation, and that such genes often play important roles in a diversity of adaptive and disease-related phenotypes in humans.

Nested protein-coding genes accumulated throughout metazoan evolution, with early analyses of human and Drosophila microarray data indicating that this phenomenon was simply due to the presence of large introns. However, a recent study employing RNA-seq data uncovered evidence of transcriptional interference driving rapid expression divergence between Drosophila nested genes, illustrating that accurate expression estimation of overlapping genes can enhance detection of their relationships. Hence, here I apply an analogous approach to strand-specific RNA-seq data from human and mouse to revisit the role of transcriptional interference in the evolution of mammalian nested genes. A genomic survey reveals that whereas mammalian nested genes indeed accrued over evolutionary time, they are retained at lower frequencies than in Drosophila. Though several properties of mammalian nested genes align with observations in Drosophila and with expectations under transcriptional interference, contrary to both, their expression divergence is not statistically different from that between unnested genes, and also does not increase after nesting. Together, these results support the hypothesis that lower selection efficiencies limit rates of gene expression evolution in mammals, leading to their reliance on immediate eradication of deleterious nested genes to avoid transcriptional interference.

In 1970, Susumu Ohno hypothesized that gene duplication was a major reservoir of adaptive innovation. However, it was not until over two decades later that DNA sequencing studies uncovered the ubiquity of gene duplication across all domains of life, highlighting its global importance in the evolution of phenotypic complexity and species diversification. Today, it seems that there are no limits to the study of evolution by gene duplication, as it has rapidly coevolved with numerous experimental and computational advances in genomics. In this perspective, we examine word stem usage in PubMed abstracts to infer how evolving discoveries and technologies have shaped the landscape of studying evolution by gene duplication, leading to a more refined understanding of its role in the emergence of novel phenotypes.

Gene turnover is a key source of adaptive variation. Yet most evolutionary studies have focused on gene duplication, dismissing gene deletion as a mechanism that simply eradicates redundancy. Here, I use genome-scale sequence and multi-tissue expression data from Drosophila melanogaster and Drosophila pseudoobscura to simultaneously assess the evolutionary outcomes of gene duplication and deletion in Drosophila. I find that gene duplication is more frequent than gene deletion in both species, indicating that it may play a more important role in Drosophila evolution. However, examination of several genic properties reveals that genes likely possess distinct functions after duplication that diverge further before deletion, suggesting that loss of redundancy cannot explain a majority of gene deletion events in Drosophila. Moreover, in addition to providing support for the well-known "out of the testis" origin of young duplicate genes, analyses of gene expression profiles uncover a preferential bias against deletion of old ovary-expressed genes. Therefore, I propose a novel "into the ovary" hypothesis for gene deletion in Drosophila, in which gene deletion may promote adaptation by salvaging genes that contribute to the evolution of female reproductive phenotypes. Under this combined "out of the testis, into the ovary" evolutionary model, gene duplication and deletion work in concert to generate and maintain a balanced repertoire of genes that promote sex-specific adaptation in Drosophila.

Poaceae (grasses) is an agriculturally important and widely distributed family of plants with extraordinary phenotypic diversity, much of which was generated under recent lineage-specific evolution. Yet, little is known about the genes and functional modules involved in the lineage-specific divergence of grasses. Here, I address this question on a genome-wide scale by applying a novel branch-based statistic of lineage-specific expression divergence, LED, to RNA-seq data from nine tissues of the wild grass Brachypodium distachyon and its domesticated relatives Oryza sativa japonica (rice) and Sorghum bicolor (sorghum). I find that LED is generally smallest in B. distachyon and largest in O. sativa japonica, which underwent domestication earlier than S. bicolor, supporting the hypothesis that domestication may increase the rate of lineage-specific expression divergence in grasses. Moreover, in all three species, LED is positively correlated with protein-coding sequence divergence and tissue specificity, and negatively correlated with network connectivity. Further analysis reveals that genes with large LED are often primarily expressed in anther, implicating lineage-specific expression divergence in the evolution of male reproductive phenotypes. Gene ontology enrichment analysis also identifies an overrepresentation of terms related to male reproduction in the two domesticated grasses, as well as to those involved in host-pathogen defense in all three species. Last, examinations of genes with the largest LED reveal that their lineage-specific expression divergence may have contributed to antimicrobial functions in B. distachyon, to enhanced adaptation and yield during domestication in O. sativa japonica, and to defense against a widespread and devastating fungal pathogen in S. bicolor. Together, these findings suggest that lineage-specific expression divergence in grasses may increase under domestication and preferentially target rapidly evolving genes involved in male reproduction, host-pathogen defense, and the origin of domesticated phenotypes.

Gene duplication is thought to play a major role in phenotypic evolution. Yet the forces involved in the functional divergence of young duplicate genes remain unclear. Here, we use population-genetic inference to elucidate the role of natural selection in the functional evolution of young duplicate genes in Drosophila melanogaster. We find that negative selection acts on young duplicates with ancestral functions, and positive selection on those with novel functions, suggesting that natural selection may determine whether and how young duplicate genes are retained. Moreover, evidence of natural selection is strongest in protein-coding regions and 3′ UTRs of young duplicates, indicating that selection may primarily target encoded proteins and regulatory sequences specific to 3′ UTRs. Further analysis reveals that natural selection acts immediately after duplication and weakens over time, possibly explaining the observed bias toward the acquisition of new functions by young, rather than old, duplicate gene copies. Last, we find an enrichment of testis-related functions in young duplicates that underwent recent positive selection, but not in young duplicates that did not undergo recent positive selection, or in old duplicates that either did or did not undergo recent positive selection. Thus, our findings reveal that natural selection is a key player in the functional evolution of young duplicate genes, acts rapidly and in a region-specific manner, and may underlie the origin of novel testis-specific phenotypes in Drosophila.

Nested genes are the most common form of protein-coding overlap in eukaryotic genomes. Previous studies have shown that nested genes accumulate rapidly over evolutionary time, typically via the insertion of short young duplicate genes into long introns. However, the evolutionary relationship between nested genes remains unclear. Here, I compare RNA-seq expression profiles of nested, proximal intra-chromosomal, intermediate intra-chromosomal, distant intra-chromosomal, and inter-chromosomal gene pairs in two Drosophila species. I find that expression profiles of nested genes are more divergent than those of any other class of genes, supporting the hypothesis that concurrent expression of nested genes is deleterious due to transcriptional interference. Further analysis reveals that expression profiles of derived nested genes are more divergent than those of their ancestral un-nested orthologs, which are more divergent than those of un-nested genes with similar genomic features. Thus, gene expression divergence between nested genes is likely caused by selection against nesting of genes with insufficiently divergent expression profiles, as well as by continued expression divergence after nesting. Moreover, expression divergence and sequence evolutionary rates are elevated in young nested genes and reduced in old nested genes, indicating that a burst of rapid evolution occurs after nesting. Together, these findings suggest that similarity between expression profiles of nested genes is deleterious due to transcriptional interference, and that natural selection addresses this problem both by eradicating highly deleterious nestings and by enabling rapid expression divergence of surviving nested genes, thereby quickly limiting or abolishing transcriptional interference.

Gene duplication is a key source of genetic innovation that plays a role in the evolution of phenotypic complexity. Although several evolutionary processes can result in the long-term retention of duplicate genes, their relative contributions in nature are unknown. Here we develop a phylogenetic approach for comparing genome-wide expression profiles of closely related species to quantify the roles of conservation, neofunctionalization, subfunctionalization, and specialization in the preservation of duplicate genes. Application of our method to pairs of young duplicates in Drosophila shows that neofunctionalization, the gain of a novel function in one copy, accounts for the retention of almost two-thirds of duplicate genes. Surprisingly, novel functions nearly always originate in younger (child) copies, whereas older (parent) copies possess functions similar to those of ancestral genes. Further examination of such pairs reveals a strong bias toward RNA-mediated duplication events, implicating asymmetric duplication and positive selection in the evolution of new functions. Moreover, we show that young duplicate genes are expressed primarily in testes and that their expression breadth increases over evolutionary time. This finding supports the "out-of-testes" hypothesis, which posits that testes are a catalyst for the emergence of new genes that ultimately evolve functions in other tissues. Thus, our study highlights the importance of neofunctionalization and positive selection in the retention of young duplicates in Drosophila and illustrates how duplicates become incorporated into novel functional networks over evolutionary time.

Sex-biased genes are thought to drive phenotypic differences between males and females. The recent availability of high-throughput gene expression data for many related species has led to a burst of investigations into the genomic and evolutionary properties of sex-biased genes. In Drosophila, a number of studies have found that X chromosomes are deficient in male-biased genes (demasculinized) and enriched for female-biased genes (feminized) and that male-biased genes evolve faster than female-biased genes. However, studies have yielded vastly different conclusions regarding the numbers of sex-biased genes and forces shaping their evolution. Here, we use RNA-seq data from multiple tissues of Drosophila melanogaster and D. pseudoobscura, a species with a recently evolved X chromosome, to explore the evolution of sex-biased genes in Drosophila. First, we compare several independent metrics for classifying sex-biased genes and find that the overlap of genes identified by different metrics is small, particularly for female-biased genes. Second, we investigate genome-wide expression patterns and uncover evidence of demasculinization and feminization of both ancestral and new X chromosomes, demonstrating that gene content on sex chromosomes evolves rapidly. Third, we examine the evolutionary rates of sex-biased genes and show that male-biased genes evolve much faster than female-biased genes, which evolve at similar rates to unbiased genes. Analysis of gene expression among tissues reveals that this trend may be partially due to pleiotropic effects of female-biased genes, which limits their evolutionary potential. Thus, our findings illustrate the importance of accurately identifying sex-biased genes and provide insight into their evolutionary dynamics in Drosophila.

Gene conversion is the unidirectional transfer of genetic information between allelic (orthologous) or nonallelic (paralogous) DNA segments. Recently, there has been much interest in understanding how gene conversion shapes the nucleotide composition of the genomic landscape. A widely held hypothesis is that gene conversion is universally GC-biased. However, direct experimental evidence of this hypothesis is limited to a single study of meiotic crossovers in yeast. Although there have been a number of indirect studies of gene conversion, evidence of GC-biased replacements gathered from such studies can also be attributed to positive selection, which has the same evolutionary dynamics as biased gene conversion. Here, we apply a direct phylogenetic approach to examine nucleotide replacements produced by nonallelic gene conversion in Drosophila and primate genomes. We find no evidence for GC-biased gene conversion in either lineage, suggesting that previously observed GC biases may be due to positive selection rather than to biased gene conversion.

New Product Portfolio Management is aimed at helping decision-makers better select projects for new products based on key criteria for the manufacturer. The Brazilian pharmaceutical industry has been undergoing change due to stricter sanitary requirements following the enactment of the Generic Law in 1999. This paper presents the results of a research study aimed at clarifying the rationale employed by national pharmaceutical companies in selecting and prioritizing their new product development projects. Consequently, proposals for an analytical structure that could help these companies better select their products were produced. The research was carried out using case study methodology in which four different companies were investigated. The results of the field study confirmed that these companies had a non-structured Product Development System and that the selection of new product development projects was made on a non-systematic basis. The research also identified key criteria for the selection of projects of new pharmaceutical products, which provided the basis for the preparation of a proposal for a managerial standard for application of New Product Portfolio Management.A gestão de portfólio de projetos de novos produtos visa a auxiliar os tomadores de decisão a selecionar projetos de novos produtos considerando critérios importantes para a organização. A indústria farmacêutica brasileira tem passado por transformações devido ao aumento das exigências sanitárias após a Lei de Genéricos, de 1999. O objetivo deste trabalho foi entender como as indústrias farmacêuticas brasileiras selecionam seus projetos de desenvolvimento de novos produtos e propor uma estrutura que possa auxiliar estas empresas a selecionar seus projetos de produtos. Foi utilizada a metodologia de estudo de caso e uma mostra de quatro organizações foi investigada. Os resultados indicam que essas empresas apresentam um desenvolvimento de produtos não estruturado e que a seleção de projetos de novos produtos é realizada de forma não-sistemática. Critérios importantes para a seleção de projetos de novos produtos foram identificados e utilizados para elaboração de um padrão gerencial para aplicação da gestão de portfólio de projetos de novos produtos.

Background Gene duplication provides raw material for the evolution of functional innovation. We recently developed a phylogenetic method that classifies evolutionary processes driving the retention of duplicate genes by quantifying divergence between their spatial gene expression profiles and that of their single-copy orthologous gene in a closely related sister species. Results Here, we apply our classification method to pairs of duplicate genes in eight mammalian genomes, using data from 11 tissues to construct spatial gene expression profiles. We find that young mammalian duplicates are often functionally conserved, and that expression divergence rapidly increases over evolutionary time. Moreover, expression divergence results in increased tissue specificity, with an overrepresentation of expression in male kidney, underrepresentation of expression in female liver, and strong underrepresentation of expression in testis. Thus, duplicate genes acquire a diversity of new tissue-specific functions outside of the testis, possibly contributing to the origin of a multitude of complex phenotypes during mammalian evolution. Conclusions Our findings reveal that mammalian duplicate genes are initially functionally conserved, and then undergo rapid functional divergence over evolutionary time, acquiring diverse tissue-specific biological roles. These observations are in stark contrast to the much faster expression divergence and acquisition of broad housekeeping roles we previously observed in Drosophila duplicate genes. Due to the smaller effective population sizes of mammals relative to Drosophila, these analyses implicate natural selection in the functional evolution of duplicate genes. Electronic supplementary material The online version of this article (doi:10.1186/s12862-015-0426-x) contains supplementary material, which is available to authorized users.

Gene duplication is thought to play a key role in phenotypic innovation. While several processes have been hypothesized to drive the retention and functional evolution of duplicate genes, their genomic contributions have never been determined. We recently developed the first genome-wide method to classify these processes by comparing distances between expression profiles of duplicate genes and their ancestral single-copy orthologs. Application of our approach to spatial gene expression profiles in two Drosophila species revealed that a majority of young duplicate genes possess new functions, and that new functions are acquired rapidly-often within a few million years. Surprisingly, new functions tend to arise in younger copies of duplicate gene pairs. Moreover, we found that young duplicates are often specifically expressed in testes, whereas old duplicates are broadly expressed across several tissues, providing strong support for the hypothetical "out-of-testes" origin of new genes. In this Extra View, I discuss our findings in the context of theoretical predictions about gene duplication, with a particular emphasis on the importance of natural selection in the evolution of novel phenotypes.

A key question in evolutionary genomics is how populations navigate the adaptive landscape in the presence of epistasis, or interactions among loci. This problem can be directly addressed by studying the evolution of RNA secondary structures, for which there is constraint to maintain pairing between Watson-Crick (WC) sites. Replacement of a nucleotide at one site of a WC pair reduces fitness by disrupting binding, which can be restored via a compensatory replacement at the interacting site. Here, I present the first genome-scale analysis of epistasis on the RNA secondary structure of human immunodeficiency virus type 1 (HIV-1). Comparison of polymorphism frequencies at ancestrally conserved sites reveals that selection against replacements is ∼2.7 times stronger at WC than at non-WC sites, such that nearly 50% of constraint can be attributed to epistasis. However, almost all epistatic constraint is due to selection against conversions of WC pairs to unpaired (UP) nucleotides, whereas conversions to GU wobbles are only slightly deleterious. This disparity is also evident in pairs with second-site compensatory replacements; conversions from UP nucleotides to WC pairs increase median fitness by ∼4.2%, whereas conversions from GU wobbles to WC pairs only increase median fitness by ∼0.3%. Moreover, second-site replacements that convert UP nucleotides to GU wobbles also increase median fitness by ∼4%, indicating that such replacements are nearly as compensatory as those that restore WC pairing. Thus, WC peaks of the HIV-1 epistatic adaptive landscape are connected by high GU ridges, enabling the viral population to rapidly explore distant peaks without traversing deep UP valleys., Author Summary Epistasis is an evolutionary process in which the effect of a nucleotide at one site in the genome is dependent on the presence or absence of particular nucleotides at other sites in the genome. One of the simplest types of epistasis occurs between Watson-Crick (WC) nucleotides in RNA secondary structures, which are under constraint to maintain base-pairing. In this study, I examine the effects of mutations at WC sites in the RNA secondary structure of HIV-1. I show that while epistasis plays a major role in the evolution of the HIV-1 secondary structure, different types of mutations have variable effects on fitness. Therefore, by favoring certain mutational trajectories, HIV-1 can evolve rapidly despite strong epistatic constraint on its RNA secondary structure.

Summary: Bridges is a heuristic search tool that uses short word matches to rapidly identify local similarities between sequences. It consists of three stages: filtering input sequences, identifying local similarities and post-processing local similarities. As input sequence data are released from memory after the filtering stage, genome-scale datasets can be efficiently compared in a single run. Bridges also includes 20 parameters, which enable the user to dictate the sensitivity and specificity of a search. Availability: Bridges is implemented in the C programming language and can be run on all platforms. Source code and documentation are available at http://github.com/rassis/bridges. Contact: kondrash@umich.edu

The most common form of protein-coding gene overlap in eukaryotes is a simple nested structure, whereby one gene is embedded in an intron of another. Analysis of nested protein-coding genes in vertebrates, fruit flies and nematodes revealed substantially higher rates of evolutionary gains than losses. The accumulation of nested gene structures could not be attributed to any obvious functional relationships between the genes involved and represents an increase of the organizational complexity of animal genomes via a neutral process.

Gene conversion is the unidirectional transfer of genetic information between orthologous (allelic) or paralogous (nonallelic) genomic segments. Though a number of studies have examined nucleotide replacements, little is known about length difference mutations produced by gene conversion. Here, we investigate insertions and deletions produced by nonallelic gene conversion in 338 Drosophila and 10,149 primate paralogs. Using a direct phylogenetic approach, we identify 179 insertions and 614 deletions in Drosophila paralogs, and 132 insertions and 455 deletions in primate paralogs. Thus, nonallelic gene conversion is strongly deletion-biased in both lineages, with almost 3.5 times as many conversion-induced deletions as insertions. In primates, the deletion bias is considerably stronger for long indels and, in both lineages, the per-site rate of gene conversion is orders of magnitudes higher than that of ordinary mutation. Due to this high rate, deletion-biased nonallelic gene conversion plays a key role in genome size evolution, leading to the cooperative shrinkage and eventual disappearance of selectively neutral paralogs., Author Summary Gene conversion is a process whereby a DNA sequence is copied from one segment of the genome (donor) to another (recipient), resulting in the replacement, insertion, or deletion of a DNA sequence in the recipient. This exchange is facilitated by the high sequence similarity of the two segments, which is due to their evolutionary relationship. Here, we study insertions and deletions produced by gene conversion between paralogs, segments related by DNA duplication events. By comparing paralog sequences in multiple species of fruit flies and primates, we find that deletions occur more than three times as frequently as insertions. We also discover that the rate of gene conversion between paralogs is quite high. The deletion bias and high rate of this process causes paralogs to shrink cooperatively and eventually be eliminated from the genome. Because of the abundance of paralogs in animal genomes, this phenomenon can lead to a significant reduction in genome size. Therefore, our finding enhances our understanding of the forces that lead to changes in genome size during evolution.