CoChat
Sign Up
Cited by 22Genetic diversity and population structureForensic and Genetic ResearchEvolution and Genetic Dynamics

This population does not exist: learning the distribution of evolutionary histories with generative adversarial networks

Open access
Booker, William W, Ray, Dylan D, Schrider, Daniel R·2023·GENETICS
DOI 10.1093/genetics/iyad063PMC PMC10213497

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Numerous studies over the last decade have demonstrated the utility of machine learning methods when applied to population genetic tasks. More recent studies show the potential of deep-learning methods in particular, which allow researchers to approach problems without making prior assumptions about how the data should be summarized or manipulated, instead learning their own internal representation of the data in an attempt to maximize inferential accuracy. One type of deep neural network, called Generative Adversarial Networks (GANs), can even be used to generate new data, and this approach has been used to create individual artificial human genomes free from privacy concerns. In this study, we further explore the application of GANs in population genetics by designing and training a network to learn the statistical distribution of population genetic alignments (i.e. data sets consisting of sequences from an entire population sample) under several diverse evolutionary histories—the first GAN capable of performing this task. After testing multiple different neural network architectures, we report the results of a fully differentiable Deep-Convolutional Wasserstein GAN with gradient penalty that is capable of generating artificial examples of population genetic alignments that successfully mimic key aspects of the training data, including the site-frequency spectrum, differentiation between populations, and patterns of linkage disequilibrium. We demonstrate consistent training success across various evolutionary models, including models of panmictic and subdivided populations, populations at equilibrium and experiencing changes in size, and populations experiencing either no selection or positive selection of various strengths, all without the need for extensive hyperparameter tuning. Overall, our findings highlight the ability of GANs to learn and mimic population genetic data and suggest future areas where this work can be applied in population genetics research that we discuss herein.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/genetics/iyad063

Cites

7 of 7 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 64

  1. Adrion, 2020, A community-maintained standard library of population genetic models, eLife, 9, 10.7554/eLife.54967 · DOI
  2. Adrion, 2020, Predicting the landscape of recombination using deep learning, Mol Biol Evol., 37, 1790, 10.1093/molbev/msaa038 · DOI
  3. Arjovsky, 2017, Wasserstein GAN, arXiv, 10.48550/arxiv.1701.07875 · DOI · Paper page
  4. Battey, 2021, Visualizing population structure with variational autoencoders, G3 (Bethesda), 11, 10.1093/g3journal/jkaa036 · DOI
  5. Baumdicker, 2022, Efficient ancestry and mutation simulation with msprime 1.0, Genetics, 220, 10.1093/genetics/iyab229 · DOI
  6. Beaumont, 2010, Approximate Bayesian computation in evolution and ecology, Annu Rev Ecol Evol Syst., 41, 379, 10.1146/annurev-ecolsys-102209-144621 · DOI
  7. Benegas, 2022, DNA language models are powerful zero-shot predictors of non-coding variant effects, BioRxiv, 10.1101/2022.08.22.504706 · DOI
  8. Bonneel, 2015, Sliced and radon Wasserstein barycenters of measures, J Math Imaging Vis, 51, 22, 10.1007/s10851-014-0506-3 · DOI
  9. Breiman, 2001, Statistical modeling: the two cultures (with comments and a rejoinder by the author), Stat Sci., 16, 199, 10.1214/ss/1009213726 · DOI · Paper page
  10. Burkart, 2021, A survey on the explainability of supervised machine learning, JAIR, 70, 245, 10.1613/jair.1.12228 · DOI · Paper page
  11. Creswell, 2018, Generative adversarial networks: an overview, IEEE Signal Process Mag, 35, 53, 10.1109/MSP.2017.2765202 · DOI · Paper page
  12. Fay, 2000, Hitchhiking under positive Darwinian selection, Genetics, 155, 1405, 10.1093/genetics/155.3.1405 · DOI · Paper page
  13. Ferrer-Admetlla, 2014, On detecting incomplete soft or hard selective sweeps using haplotype structure, Mol Biol Evol., 31, 1275, 10.1093/molbev/msu077 · DOI
  14. Flagel, 2019, The unreasonable effectiveness of convolutional neural networks in population genetic inference, Mol Biol Evol., 36, 220, 10.1093/molbev/msy224 · DOI
  15. Goodfellow, 2020, Generative adversarial networks, Commun ACM., 63, 139, 10.48550/arxiv.1406.2661 · DOI · Paper page
  16. Gower, 2021, Detecting adaptive introgression in human evolution using convolutional neural networks, eLife, 10, e64669, 10.7554/eLife.64669 · DOI · Paper page
  17. Gulrajani, [ 6 2023] 2017, Improved training of Wasserstein GANs, arXiv, 10.48550/arXiv.1704.00028 · DOI · Paper page
  18. Harris, 2020, Array programming with NumPy, Nature, 585, 357, 10.1038/s41586-020-2649-2 · DOI · Paper page
  19. Hudson, 2002, Generating samples under a Wright-Fisher neutral model of genetic variation, Bioinformatics, 18, 337, 10.1093/bioinformatics/18.2.337 · DOI
  20. Hudson, 1995, Deleterious background selection with recombination, Genetics, 141, 1605, 10.1093/genetics/141.4.1605 · DOI
  21. Isildak, 2021, Distinguishing between recent balancing selection and incomplete sweep using deep neural networks, Mol Ecol Resour., 21, 2706, 10.1111/1755-0998.13379 · DOI
  22. Jang, 2016, Categorical reparameterization with gumbel-softmax, arXiv, 10.48550/arxiv.1611.01144 · DOI · Paper page
  23. Jensen, 2005, Distinguishing between selective sweeps and demography using DNA polymorphism data, Genetics, 170, 1401, 10.1534/genetics.104.038224 · DOI
  24. Kaplan, 1989, The “hitchhiking effect” revisited, Genetics, 123, 887, 10.1093/genetics/123.4.887 · DOI · Paper page
  25. Karras, [ 6 2023] 2021, Alias-free generative adversarial networks, arXiv, 10.48550/arXiv.2106.12423 · DOI
  26. Karras, 2018, A style-based generator architecture for generative adversarial networks, arXiv, 10.48550/arxiv.1812.04948 · DOI
  27. Kelleher, 2016, Efficient coalescent simulation and genealogical analysis for large sample sizes, PLoS Comput Biol., 12, e1004842, 10.1371/journal.pcbi.1004842 · DOI
  28. Kern, 2016, Discoal: flexible coalescent simulations with selection, Bioinformatics, 32, 3839, 10.1093/bioinformatics/btw556 · DOI
  29. Kern, 2018, Diplos/HIC: an updated approach to classifying selective sweeps, G3 (Bethesda), 8, 1959, 10.1534/g3.118.200262 · DOI
  30. Khan, 2022, scAEGAN: unification of single-cell genomics data by adversarial learning of latent space correspondences, BioRxiv, 10.1101/2022.04.19.488745 · DOI
  31. Kim, 2021, Contrastive fine-grained class clustering via generative adversarial networks, arXiv, 10.48550/arxiv.2112.14971 · DOI
  32. Kim, 2004, Linkage disequilibrium as a signature of selective sweeps, Genetics, 167, 1513, 10.1534/genetics.103.025387 · DOI
  33. Kingma, 2014, Adam: a method for stochastic optimization, arXiv, 10.48550/arxiv.1412.6980 · DOI · Paper page
  34. LeCun, 1989, Backpropagation applied to handwritten zip code recognition, Neural Comput, 1, 541, 10.1162/neco.1989.1.4.541 · DOI · Paper page
  35. Li, 2011, Inference of human population history from individual whole-genome sequences, Nature, 475, 493, 10.1038/nature10231 · DOI · Paper page
  36. Liu, 2022, Towards disentangling latent space for unsupervised semantic face editing, IEEE Trans Image Process, 31, 1475, 10.1109/TIP.2022.3142527 · DOI
  37. Liu, 2015, Exploring population size changes using SNP frequency spectra, Nat Genet., 47, 555, 10.1038/ng.3254 · DOI
  38. Liu, 2020, Stairway plot 2: demographic history inference with folded SNP frequency spectra, Genome Biol, 21, 280, 10.1186/s13059-020-02196-9 · DOI
  39. Maddison, 2016, The concrete distribution: a continuous relaxation of discrete random variables, arXiv, 10.48550/arxiv.1611.00712 · DOI
  40. Miles, 2021, cggh/scikit-allel: v1.3.3
  41. Nei, 1979, Mathematical model for studying genetic variation in terms of restriction endonucleases, Proc Natl Acad Sci USA, 76, 5269, 10.1073/pnas.76.10.5269 · DOI
  42. Nordborg, 1996, The effect of recombination on background selection, Genet Res., 67, 159, 10.1017/s0016672300033619 · DOI
  43. Paszke, [ 6 2023] 2019, arXiv, 10.48550/arXiv.1912.01703 · DOI · Paper page
  44. Pybus, 2015, Hierarchical boosting: a machine-learning framework to detect and classify hard selective sweeps in human populations, Bioinformatics, 31, 3946, 10.1093/bioinformatics/btv493 · DOI
  45. Radford, 2015, Unsupervised representation learning with deep convolutional generative adversarial networks, arXiv, 10.48550/arxiv.1511.06434 · DOI · Paper page
  46. Salimans, [ 6 2023] 2016, Improved techniques for training GANs, arXiv, 10.48550/arXiv.1606.03498 · DOI · Paper page
  47. Sanchez, 2021, Deep learning for population size history inference: design, comparison and combination with approximate Bayesian computation, Mol Ecol Resour., 21, 2645, 10.1111/1755-0998.13224 · DOI
  48. Schiffels, 2014, Inferring human population size and separation history from multiple genome sequences, Nat Genet., 46, 919, 10.1038/ng.3015 · DOI · Paper page
  49. Schrider, 2015, Inferring selective constraint from population genomic data suggests recent regulatory turnover in the human brain, Genome Biol Evol, 7, 3511, 10.1093/gbe/evv228 · DOI
  50. Schrider, 2016, S/HIC: robust identification of soft and hard sweeps using machine learning, PLoS Genet, 12, e1005928, 10.1371/journal.pgen.1005928 · DOI
  51. Simonsen, 1995, Properties of statistical tests of neutrality for DNA polymorphism data, Genetics, 141, 413, 10.1093/genetics/141.1.413 · DOI · Paper page
  52. Simonyan, 2013, Deep inside convolutional networks: visualising image classification models and saliency maps, arXiv, 10.48550/arxiv.1312.6034 · DOI · Paper page
  53. Smith, 1974, The hitch-hiking effect of a favourable gene, Genet Res., 23, 23, 10.1017/S0016672300014634 · DOI · Paper page
  54. Smith, 2022, Dispersal inference from population genetic variation using a convolutional neural network, BioRxiv, 10.1101/2022.08.25.505329 · DOI
  55. Stephan, 2016, Signatures of positive selection: from selective sweeps at individual loci to subtle allele frequency changes in polygenic adaptation, Mol Ecol., 25, 79, 10.1111/mec.13288 · DOI
  56. Tajima, 1989, Statistical method for testing the neutral mutation hypothesis by DNA polymorphism, Genetics, 123, 585, 10.1093/genetics/123.3.585 · DOI
  57. Terhorst, 2017, Robust and scalable inference of population history from hundreds of unphased whole genomes, Nat Genet., 49, 303, 10.1038/ng.3748 · DOI
  58. Voight, 2006, A map of recent positive selection in the human genome, PLoS Biol, 4, e72, 10.1371/journal.pbio.0040072 · DOI
  59. Wang, 2020, Automatic inference of demographic parameters using generative adversarial networks, BioRxiv, 10.1101/2020.08.05.237834 · DOI
  60. Watterson, 1975, On the number of segregating sites in genetical models without recombination, Theor Popul Biol., 7, 256, 10.1016/0040-5809(75)90020-9 · DOI
  61. Whitehouse, 2022, Timesweeper: accurately identifying selective sweeps using population genomic time series, BioRxiv, 10.1101/2022.07.06.499052 · DOI
  62. Yale, 2019, Privacy Preserving Synthetic Health Data, in hal-02160496
  63. Yelmen, 2021, Creating artificial human genomes using generative neural networks, PLoS Genet, 17, 10.1371/journal.pgen.1009303 · DOI
  64. Zhu, 2017, Unpaired image-to-image translation using cycle-consistent adversarial networks, arXiv, 10.48550/arxiv.1703.10593 · DOI

Authors
Booker, W. W., Ray, D. D., & Schrider, D. R.
Affiliations
University of North Carolina at Chapel Hill (US)
Corresponding author
  • Booker, William W
  • Schrider, Daniel R
ORCID
Venue
GENETICS
Publisher
Oxford University Press (OUP)
Published
2023-04-17
Volume
224
Issue
2
ISSN
  • 0016-6731 (ISSN-L)
  • 1943-2631
DOI
10.1093/genetics/iyad063
PMC
PMC10213497
OpenAlex
W4366083659
Cited by
22
References
7
Citations by year
2026: 5 · 2025: 5 · 2024: 7 · 2023: 5
FWCI
3.65
Type
journal article
Language
en
Open access
Licence
academic.oup.com/pages/standard-publication-reuse-rights (vor)
Locations
  • Genetics (publishedVersion)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc)
  • PubMed Central (publishedVersion)
  • PubMed (publishedVersion)
Funders
NIH (HG010774, GM138286)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic diversity and population structure
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Forensic and Genetic Research
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Evolution and Genetic Dynamics
Keywords
generative adversarial networks, synthetic data generation, site frequency spectrum, linkage disequilibrium, genetic differentiation, positive selection, population size change, panmictic population, subpopulations
MeSH terms
Machine Learning, Biological Evolution, Humans, Linkage Disequilibrium, Genome, Human, Privacy

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review