Molly Przeworski
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Molly Przeworski has a total of 170 co-authors. Here are the top co-authors by number of shared publications:
- Ziyue Gao
Affiliation Department of Genetics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America
8 shared publications - Guy Sella
Department of Biological Sciences, Columbia University, New York, New York, United States of America
7 shared publications - Ipsita Agarwal
Department of Biological Sciences, Columbia University
7 shared publications - Marc de Manuel
Affiliation Institute of Evolutionary Biology, CSIC-Universitat Pompeu Fabra, Barcelona, Spain
7 shared publications - Zachary Baker
Department of Systems Biology, Columbia University
7 shared publications - Zachary L. Fuller
Department of Biological Sciences, Columbia University, New York, NY, USA
7 shared publications
Latest Publications
- Open Access
Dynamics of mutators of arbitrary dominance in humans
Recent findings in humans and other species have revealed the presence of "mutator" alleles that increase germline mutation rate across the genome. Such mutators are expected to be selected against because of the additional deleterious alleles that they generate, to a degree that will depend on how much they increase the mutation rate in heterozygotes and homozygotes. To describe their dynamics, we develop a population genetic model of mutation rate modifiers with arbitrary dominance coefficients, in which fitness effects stem from additional germline mutations. We then use it to interpret findings for the seven human mutators identified to date. For six of the seven known mutators, the observed frequencies are well fit by the model and thus consistent with purifying selection arising solely due to their effects on germline mutation rates, although also consistent with a wide range of parameters; in two, the observed frequencies are readily explained by purely recessive fitness effects. The exception is a variant in MUTYH , which is more common than predicted under plausible parameters, for reasons that remain unclear. We also use the model to explore what types of mutators are most likely to be discovered in parents, based on identifying offspring with unexpectedly high numbers of de novo mutations. Although the two first mutators identified by this approach seem to be recessive, our modeling suggests that, for the same effect size, semi-dominant mutators are much more likely to be detected. These findings therefore imply that there are many more modifier sites with recessive effects than semi-dominant ones. More generally, our model provides a framework for interpreting properties of mutators in humans and other species and for learning about the genetic architecture of germline mutation rate variation.
Open AccessDynamics of mutators of arbitrary dominance in humans
Recent findings in humans and other species have revealed the presence of "mutator" alleles that increase germline mutation rate across the genome. Such mutators are expected to be selected against because of the additional deleterious alleles that they generate, to a degree that will depend on how much they increase the mutation rate in heterozygotes and homozygotes. To describe their dynamics, we develop a population genetic model of mutation rate modifiers with arbitrary dominance coefficients, in which fitness effects stem from additional germline mutations. We then use it to interpret findings for the seven human mutators identified to date. For six of the seven known mutators, the observed frequencies are well fit by the model and thus consistent with purifying selection arising solely due to their effects on germline mutation rates, although also consistent with a wide range of parameters; in two, the observed frequencies are readily explained by purely recessive fitness effects. The exception is a variant in MUTYH , which is more common than predicted under plausible parameters, for reasons that remain unclear. We also use the model to explore what types of mutators are most likely to be discovered in parents, based on identifying offspring with unexpectedly high numbers of de novo mutations. Although the two first mutators identified by this approach seem to be recessive, our modeling suggests that, for the same effect size, semi-dominant mutators are much more likely to be detected. These findings therefore imply that there are many more modifier sites with recessive effects than semi-dominant ones. More generally, our model provides a framework for interpreting properties of mutators in humans and other species and for learning about the genetic architecture of germline mutation rate variation.
- Open Access
- 8 June 2026
- Published by Public Library of Science (PLoS) in PLOS Biology
- Vol. 24 (6) , e3003799
What sets the mutation rate of a cell type in an animal species?
Germline mutation rates per generation are strikingly similar across animals, despite vast differences in life histories. Analogously, in at least one somatic cell type, mutation burdens at the end of lifespans are comparable across mammals. These observations point to a key role for natural selection in shaping mutation rates. This Essay summarizes the patterns identified to date and outlines existing theories for how selection pressures might shape mutation rates in animal germline and soma. An understanding of what sets the mutation rate of a given cell type in a species requires better integration of genetics and development with population processes of selection and genetic drift.
- 8 June 2026
- Published by Public Library of Science (PLoS) in PLOS Biology
- Vol. 24 (6) , e3003799
Open AccessWhat sets the mutation rate of a cell type in an animal species?
Germline mutation rates per generation are strikingly similar across animals, despite vast differences in life histories. Analogously, in at least one somatic cell type, mutation burdens at the end of lifespans are comparable across mammals. These observations point to a key role for natural selection in shaping mutation rates. This Essay summarizes the patterns identified to date and outlines existing theories for how selection pressures might shape mutation rates in animal germline and soma. An understanding of what sets the mutation rate of a given cell type in a species requires better integration of genetics and development with population processes of selection and genetic drift.
- Open Access
A sibling study of variation in parental mutation rates
People are born with variable numbers of de novo germline mutations (DNMs), depending primarily on the ages of their parents. To explore additional causes, we developed an approach to call DNMs from nucleotide differences between siblings in genomic regions inherited identical by descent from both parents. Applying it to whole genome sequences from 28,985 sibling pairs of diverse genetic ancestries present in the UK Biobank and All of Us datasets, as well as 2,330 trios, we identified > 800K autosomal DNMs and characterized mutation phenotypes in 27,645 sets of parents. We found subtle shifts in the mutation spectrum but no differences in total DNM rates among genetic ancestry groups, or between smokers and non-smokers. Testing for associations between parental mutation phenotypes and their burden of loss-of-function and deleterious missense variants in a set of 180 DNA repair and maintenance genes, we discovered that disruptions in REV1 and LIG1 increase germline mutation rates, and thus that rare mutator alleles segregate in population cohorts.
Open AccessA sibling study of variation in parental mutation rates
People are born with variable numbers of de novo germline mutations (DNMs), depending primarily on the ages of their parents. To explore additional causes, we developed an approach to call DNMs from nucleotide differences between siblings in genomic regions inherited identical by descent from both parents. Applying it to whole genome sequences from 28,985 sibling pairs of diverse genetic ancestries present in the UK Biobank and All of Us datasets, as well as 2,330 trios, we identified > 800K autosomal DNMs and characterized mutation phenotypes in 27,645 sets of parents. We found subtle shifts in the mutation spectrum but no differences in total DNM rates among genetic ancestry groups, or between smokers and non-smokers. Testing for associations between parental mutation phenotypes and their burden of loss-of-function and deleterious missense variants in a set of 180 DNA repair and maintenance genes, we discovered that disruptions in REV1 and LIG1 increase germline mutation rates, and thus that rare mutator alleles segregate in population cohorts.
- Open Access
What sets the mutation rate of a cell type in an animal species?
Germline mutation rates per generation are strikingly similar across animals, despite vast differences in life histories. Analogously, in at least one somatic cell type, mutation rates at the end of lifespan are comparable across mammals. These observations point to a key role for natural selection in shaping mutation rates. In this essay, we summarize the patterns identified to date and outline existing theories for how selection pressures might shape mutation rates in animal germline and soma. We argue that an understanding of what sets the mutation rate of a given cell type in a species requires better integration of genetics and development with population processes of selection and genetic drift.
Open AccessWhat sets the mutation rate of a cell type in an animal species?
Germline mutation rates per generation are strikingly similar across animals, despite vast differences in life histories. Analogously, in at least one somatic cell type, mutation rates at the end of lifespan are comparable across mammals. These observations point to a key role for natural selection in shaping mutation rates. In this essay, we summarize the patterns identified to date and outline existing theories for how selection pressures might shape mutation rates in animal germline and soma. We argue that an understanding of what sets the mutation rate of a given cell type in a species requires better integration of genetics and development with population processes of selection and genetic drift.
- Open Access
Revisiting the evidence for long-lived balancing selection in humans
Balancing selection maintains variation in a population longer than expected under neutrality. In humans, there are dozens of tentative candidate loci for balancing selection, but only a handful of well-characterized examples, which are either evolutionarily recent alleles or ancient variants shared across species identical by descent (“trans-species polymorphisms”). Here, we look for evidence of balancing selection over a range of timescales, by taking an approach that does not rely on a demographic model or assumptions about the specific mode of balancing selection. Analyzing whole genome sequencing data from 2504 humans and 59 chimpanzees, we identify common single nucleotide polymorphisms (SNPs) that are identical in the two species. This set includes recurrent mutations, a subset of which may be maintained by balancing selection in one or both species, as well as potential trans-species polymorphisms. Using allele ages estimated from ancestral recombination graph reconstructions in humans, we show that shared SNPs are enriched for older alleles as compared to matched human SNPs that are not shared with chimpanzees. On this basis, we estimate that balancing selection has maintained over one thousand alleles in humans longer than expected by chance. Moreover, we identify over 50 trans-species polymorphisms, including an intriguing case that includes an eQTL for the gene MUC7 . However, we also estimate a minimum false discovery rate for any allele age cut-off of ∼70%; as we show, even among the trans-species polymorphisms, many may be shared between humans and chimpanzees simply by chance. Thus, while our empirical approach establishes that there are numerous loci under balancing selection yet to be found, the specific targets remain difficult to identify without independent lines of evidence.
Open AccessRevisiting the evidence for long-lived balancing selection in humans
Balancing selection maintains variation in a population longer than expected under neutrality. In humans, there are dozens of tentative candidate loci for balancing selection, but only a handful of well-characterized examples, which are either evolutionarily recent alleles or ancient variants shared across species identical by descent (“trans-species polymorphisms”). Here, we look for evidence of balancing selection over a range of timescales, by taking an approach that does not rely on a demographic model or assumptions about the specific mode of balancing selection. Analyzing whole genome sequencing data from 2504 humans and 59 chimpanzees, we identify common single nucleotide polymorphisms (SNPs) that are identical in the two species. This set includes recurrent mutations, a subset of which may be maintained by balancing selection in one or both species, as well as potential trans-species polymorphisms. Using allele ages estimated from ancestral recombination graph reconstructions in humans, we show that shared SNPs are enriched for older alleles as compared to matched human SNPs that are not shared with chimpanzees. On this basis, we estimate that balancing selection has maintained over one thousand alleles in humans longer than expected by chance. Moreover, we identify over 50 trans-species polymorphisms, including an intriguing case that includes an eQTL for the gene MUC7 . However, we also estimate a minimum false discovery rate for any allele age cut-off of ∼70%; as we show, even among the trans-species polymorphisms, many may be shared between humans and chimpanzees simply by chance. Thus, while our empirical approach establishes that there are numerous loci under balancing selection yet to be found, the specific targets remain difficult to identify without independent lines of evidence.
- Open Access
Collateral mutagenesis funnels multiple sources of DNA damage into a ubiquitous mutational signature
Mutations reflect the net effects of myriad types of damage, replication errors, and repair mechanisms, and are therefore expected to differ across cell types with distinct exposures to mutagens, division rates, and cellular programs. Yet when mutations in humans are decomposed into a set of “signatures”, one single base substitution signature, SBS5, is present across cell types and tissues, and predominates in post-mitotic neurons as well as male and female germlines [1–3]. The etiology of SBS5 is unknown. By modeling the processes by which mutations arise, we infer that SBS5 is the footprint of errors in DNA synthesis triggered by distinct types of DNA damage. Supporting this hypothesis, we find that SBS5 rates increase with signatures of endogenous and exogenous DNA damage in cancerous and non-cancerous cells and co-vary with repair rates along the genome, as expected from model predictions. These analyses indicate that SBS5 captures the output of a “funnel”, through which multiple sources of damage result in a similar mutation spectrum. As we further show, SBS5 mutations arise not only from translesion synthesis but also from DNA repair, suggesting that both mechanisms occasionally trigger the same mutagenic process.
Open AccessCollateral mutagenesis funnels multiple sources of DNA damage into a ubiquitous mutational signature
Mutations reflect the net effects of myriad types of damage, replication errors, and repair mechanisms, and are therefore expected to differ across cell types with distinct exposures to mutagens, division rates, and cellular programs. Yet when mutations in humans are decomposed into a set of “signatures”, one single base substitution signature, SBS5, is present across cell types and tissues, and predominates in post-mitotic neurons as well as male and female germlines [1–3]. The etiology of SBS5 is unknown. By modeling the processes by which mutations arise, we infer that SBS5 is the footprint of errors in DNA synthesis triggered by distinct types of DNA damage. Supporting this hypothesis, we find that SBS5 rates increase with signatures of endogenous and exogenous DNA damage in cancerous and non-cancerous cells and co-vary with repair rates along the genome, as expected from model predictions. These analyses indicate that SBS5 captures the output of a “funnel”, through which multiple sources of damage result in a similar mutation spectrum. As we further show, SBS5 mutations arise not only from translesion synthesis but also from DNA repair, suggesting that both mechanisms occasionally trigger the same mutagenic process.
- Open Access
- 23 May 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (21)
The distribution of highly deleterious variants across human ancestry groups
A major focus of human genetics is to map severe disease mutations. Increasingly, that goal is understood as requiring huge numbers of people to be sequenced from every broadly defined genetic ancestry group, so as not to miss “ancestry-specific variants.” Here, we consider whether this focus is warranted. We start from first principles considerations, based on models of mutation–drift-selection balance, which suggest that since severe disease mutations tend to be strongly deleterious, and thus evolutionarily young, they will be kept at relatively constant frequency through recurrent mutation. Therefore, highly pathogenic alleles should be shared identically by descent within extended families, not broad ancestry groups, and sequencing more people should yield similar numbers regardless of ancestry. We test the model predictions using gnomAD genetic ancestry groupings and show that they provide a good fit to the classes of variants most likely to be highly pathogenic, notably sets of loss of function alleles at strongly constrained genes. These findings clarify that strongly deleterious alleles will be found at comparable rates in people of all ancestries, and the information they provide about human biology is shared across ancestries.
- 23 May 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (21)
Open AccessThe distribution of highly deleterious variants across human ancestry groups
A major focus of human genetics is to map severe disease mutations. Increasingly, that goal is understood as requiring huge numbers of people to be sequenced from every broadly defined genetic ancestry group, so as not to miss “ancestry-specific variants.” Here, we consider whether this focus is warranted. We start from first principles considerations, based on models of mutation–drift-selection balance, which suggest that since severe disease mutations tend to be strongly deleterious, and thus evolutionarily young, they will be kept at relatively constant frequency through recurrent mutation. Therefore, highly pathogenic alleles should be shared identically by descent within extended families, not broad ancestry groups, and sequencing more people should yield similar numbers regardless of ancestry. We test the model predictions using gnomAD genetic ancestry groupings and show that they provide a good fit to the classes of variants most likely to be highly pathogenic, notably sets of loss of function alleles at strongly constrained genes. These findings clarify that strongly deleterious alleles will be found at comparable rates in people of all ancestries, and the information they provide about human biology is shared across ancestries.
- Open Access
- 15 April 2025
- Published by Public Library of Science (PLoS) in PLOS Genetics
- Vol. 21 (4) , e1011661
Germline mutation rates and fine-scale recombination parameters in zebra finch
Most of our understanding of the fundamental processes of mutation and recombination stems from a handful of disparate model organisms and pedigree studies of mammals, with little known about other vertebrates. To gain a broader comparative perspective, we focused on the zebra finch (Taeniopygia castanotis), which, like other birds, differs from mammals in its karyotype (which includes many micro-chromosomes), in the mechanism by which recombination is directed to the genome, and in aspects of ontogenesis. We collected genome sequences from three generation pedigrees that provide information about 80 meioses, inferring 202 single-point de novo mutations, 1,088 crossovers, and 275 non-crossovers. On that basis, we estimated a sex-averaged mutation rate of 5.0 × 10-9 per base pair per generation, on par with mammals that have a similar generation time (~2–3 years). Also as in mammals, we found a paternal germline mutation bias at later stages of gametogenesis (of 1.7:1) but no discernible difference between sexes in early development. Examining recombination patterns, we found that the sex-averaged crossover rate on macro-chromosomes is 0.93 cM/Mb, with a pronounced enrichment of crossovers near telomeres. In contrast, non-crossover rates are more uniformly distributed. On micro-chromosomes, sex-averaged crossover rates are substantially higher (3.96 cM/Mb), in accordance with crossover homeostasis, and both crossover and non-crossover events are more uniformly distributed. At a finer scale, recombination events overlap CpG islands more often than expected by chance, as expected in the absence of PRDM9. Estimates of the degree of GC-biased gene conversion (59%), the mean non-crossover conversion tract length (~32 bp), and the non-crossover-to-crossover ratio (5.4:1) are all comparable to those reported in primates and mice. Therefore, properties of germline mutation and recombination resolutions remain similar over large phylogenetic distances.
- 15 April 2025
- Published by Public Library of Science (PLoS) in PLOS Genetics
- Vol. 21 (4) , e1011661
Open AccessGermline mutation rates and fine-scale recombination parameters in zebra finch
Most of our understanding of the fundamental processes of mutation and recombination stems from a handful of disparate model organisms and pedigree studies of mammals, with little known about other vertebrates. To gain a broader comparative perspective, we focused on the zebra finch (Taeniopygia castanotis), which, like other birds, differs from mammals in its karyotype (which includes many micro-chromosomes), in the mechanism by which recombination is directed to the genome, and in aspects of ontogenesis. We collected genome sequences from three generation pedigrees that provide information about 80 meioses, inferring 202 single-point de novo mutations, 1,088 crossovers, and 275 non-crossovers. On that basis, we estimated a sex-averaged mutation rate of 5.0 × 10-9 per base pair per generation, on par with mammals that have a similar generation time (~2–3 years). Also as in mammals, we found a paternal germline mutation bias at later stages of gametogenesis (of 1.7:1) but no discernible difference between sexes in early development. Examining recombination patterns, we found that the sex-averaged crossover rate on macro-chromosomes is 0.93 cM/Mb, with a pronounced enrichment of crossovers near telomeres. In contrast, non-crossover rates are more uniformly distributed. On micro-chromosomes, sex-averaged crossover rates are substantially higher (3.96 cM/Mb), in accordance with crossover homeostasis, and both crossover and non-crossover events are more uniformly distributed. At a finer scale, recombination events overlap CpG islands more often than expected by chance, as expected in the absence of PRDM9. Estimates of the degree of GC-biased gene conversion (59%), the mean non-crossover conversion tract length (~32 bp), and the non-crossover-to-crossover ratio (5.4:1) are all comparable to those reported in primates and mice. Therefore, properties of germline mutation and recombination resolutions remain similar over large phylogenetic distances.
- Open Access
The distribution of highly deleterious variants across human ancestry groups
A major focus of human genetics is to map severe disease mutations. Increasingly that goal is understood as requiring huge numbers of people to be sequenced from every broadly-defined genetic ancestry group, so as not to miss “ancestry-specific variants.” Here, we argue that this focus is unwarranted. We start with first principles considerations, based on models of mutation-drift-selection balance, which suggest highly pathogenic mutations should be at similarly low frequencies across ancestry groups. Severe disease mutations tend to be strongly deleterious, and thus evolutionarily young, and are kept at relatively constant frequency through recurrent mutation. Therefore, highly pathogenic alleles are shared identical by descent within extended families, not broad ancestry groups, and sequencing more people should yield similar numbers regardless of ancestry. We illustrate these points using gnomAD genetic ancestry groupings, and show that the classes of variants most likely to be highly pathogenic, notably sets of loss of function alleles at strongly constrained genes, conform well to these predictions. While there are many important reasons to diversify genomic research, strongly deleterious alleles will be found at comparable rates in people of all ancestries, and the information they provide about human biology is shared across ancestries.
Open AccessThe distribution of highly deleterious variants across human ancestry groups
A major focus of human genetics is to map severe disease mutations. Increasingly that goal is understood as requiring huge numbers of people to be sequenced from every broadly-defined genetic ancestry group, so as not to miss “ancestry-specific variants.” Here, we argue that this focus is unwarranted. We start with first principles considerations, based on models of mutation-drift-selection balance, which suggest highly pathogenic mutations should be at similarly low frequencies across ancestry groups. Severe disease mutations tend to be strongly deleterious, and thus evolutionarily young, and are kept at relatively constant frequency through recurrent mutation. Therefore, highly pathogenic alleles are shared identical by descent within extended families, not broad ancestry groups, and sequencing more people should yield similar numbers regardless of ancestry. We illustrate these points using gnomAD genetic ancestry groupings, and show that the classes of variants most likely to be highly pathogenic, notably sets of loss of function alleles at strongly constrained genes, conform well to these predictions. While there are many important reasons to diversify genomic research, strongly deleterious alleles will be found at comparable rates in people of all ancestries, and the information they provide about human biology is shared across ancestries.
- Open Access
- 13 September 2024
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 121 (38)
Causal interpretations of family GWAS in the presence of heterogeneous effects
Family-based genome-wide association studies (GWASs) are often claimed to provide an unbiased estimate of the average causal effects (or average treatment effects; ATEs) of alleles, on the basis of an analogy between the random transmission of alleles from parents to children and a randomized controlled trial. We show that this claim does not hold in general. Because Mendelian segregation only randomizes alleles among children of heterozygotes, the effects of alleles in the children of homozygotes are not observable. This feature will matter if an allele has different average effects in the children of homozygotes and heterozygotes, as can arise in the presence of gene-by-environment interactions, gene-by-gene interactions, or differences in linkage disequilibrium patterns. At a single locus, family-based GWAS can be thought of as providing an unbiased estimate of the average effect in the children of heterozygotes (i.e., a local average treatment effect; LATE). This interpretation does not extend to polygenic scores (PGSs), however, because different sets of SNPs are heterozygous in each family. Therefore, other than under specific conditions, the within-family regression slope of a PGS cannot be assumed to provide an unbiased estimate of the LATE for any subset or weighted average of families. In practice, the potential biases of a family-based GWAS are likely smaller than those that can arise from confounding in a standard, population-based GWAS, and so family studies remain important for the dissection of genetic contributions to phenotypic variation. Nonetheless, their causal interpretation is less straightforward than has been widely appreciated.
- 13 September 2024
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 121 (38)
Open AccessCausal interpretations of family GWAS in the presence of heterogeneous effects
Family-based genome-wide association studies (GWASs) are often claimed to provide an unbiased estimate of the average causal effects (or average treatment effects; ATEs) of alleles, on the basis of an analogy between the random transmission of alleles from parents to children and a randomized controlled trial. We show that this claim does not hold in general. Because Mendelian segregation only randomizes alleles among children of heterozygotes, the effects of alleles in the children of homozygotes are not observable. This feature will matter if an allele has different average effects in the children of homozygotes and heterozygotes, as can arise in the presence of gene-by-environment interactions, gene-by-gene interactions, or differences in linkage disequilibrium patterns. At a single locus, family-based GWAS can be thought of as providing an unbiased estimate of the average effect in the children of heterozygotes (i.e., a local average treatment effect; LATE). This interpretation does not extend to polygenic scores (PGSs), however, because different sets of SNPs are heterozygous in each family. Therefore, other than under specific conditions, the within-family regression slope of a PGS cannot be assumed to provide an unbiased estimate of the LATE for any subset or weighted average of families. In practice, the potential biases of a family-based GWAS are likely smaller than those that can arise from confounding in a standard, population-based GWAS, and so family studies remain important for the dissection of genetic contributions to phenotypic variation. Nonetheless, their causal interpretation is less straightforward than has been widely appreciated.