Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data.
Investigators have recently introduced powerful methods for population genetic inference that rely on supervised machine learning from simulated data. Despite their performance advantages, these methods can fail when the simulated training data does not adequately resemble data from the real world....
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2023-11-01
|
| Series: | PLoS Genetics |
| Online Access: | https://journals.plos.org/plosgenetics/article/file?id=10.1371/journal.pgen.1011032&type=printable |
| _version_ | 1797394711147708416 |
|---|---|
| author | Ziyi Mo Adam Siepel |
| author_facet | Ziyi Mo Adam Siepel |
| author_sort | Ziyi Mo |
| collection | DOAJ |
| description | Investigators have recently introduced powerful methods for population genetic inference that rely on supervised machine learning from simulated data. Despite their performance advantages, these methods can fail when the simulated training data does not adequately resemble data from the real world. Here, we show that this "simulation mis-specification" problem can be framed as a "domain adaptation" problem, where a model learned from one data distribution is applied to a dataset drawn from a different distribution. By applying an established domain-adaptation technique based on a gradient reversal layer (GRL), originally introduced for image classification, we show that the effects of simulation mis-specification can be substantially mitigated. We focus our analysis on two state-of-the-art deep-learning population genetic methods-SIA, which infers positive selection from features of the ancestral recombination graph (ARG), and ReLERNN, which infers recombination rates from genotype matrices. In the case of SIA, the domain adaptive framework also compensates for ARG inference error. Using the domain-adaptive SIA (dadaSIA) model, we estimate improved selection coefficients at selected loci in the 1000 Genomes CEU population. We anticipate that domain adaptation will prove to be widely applicable in the growing use of supervised machine learning in population genetics. |
| first_indexed | 2024-03-09T00:23:42Z |
| format | Article |
| id | doaj.art-d1ab79f501334ce198a8c69df25996a5 |
| institution | Directory Open Access Journal |
| issn | 1553-7390 1553-7404 |
| language | English |
| last_indexed | 2024-03-09T00:23:42Z |
| publishDate | 2023-11-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS Genetics |
| spelling | doaj.art-d1ab79f501334ce198a8c69df25996a52023-12-12T05:32:57ZengPublic Library of Science (PLoS)PLoS Genetics1553-73901553-74042023-11-011911e101103210.1371/journal.pgen.1011032Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data.Ziyi MoAdam SiepelInvestigators have recently introduced powerful methods for population genetic inference that rely on supervised machine learning from simulated data. Despite their performance advantages, these methods can fail when the simulated training data does not adequately resemble data from the real world. Here, we show that this "simulation mis-specification" problem can be framed as a "domain adaptation" problem, where a model learned from one data distribution is applied to a dataset drawn from a different distribution. By applying an established domain-adaptation technique based on a gradient reversal layer (GRL), originally introduced for image classification, we show that the effects of simulation mis-specification can be substantially mitigated. We focus our analysis on two state-of-the-art deep-learning population genetic methods-SIA, which infers positive selection from features of the ancestral recombination graph (ARG), and ReLERNN, which infers recombination rates from genotype matrices. In the case of SIA, the domain adaptive framework also compensates for ARG inference error. Using the domain-adaptive SIA (dadaSIA) model, we estimate improved selection coefficients at selected loci in the 1000 Genomes CEU population. We anticipate that domain adaptation will prove to be widely applicable in the growing use of supervised machine learning in population genetics.https://journals.plos.org/plosgenetics/article/file?id=10.1371/journal.pgen.1011032&type=printable |
| spellingShingle | Ziyi Mo Adam Siepel Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. PLoS Genetics |
| title | Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. |
| title_full | Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. |
| title_fullStr | Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. |
| title_full_unstemmed | Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. |
| title_short | Domain-adaptive neural networks improve supervised machine learning based on simulated population genetic data. |
| title_sort | domain adaptive neural networks improve supervised machine learning based on simulated population genetic data |
| url | https://journals.plos.org/plosgenetics/article/file?id=10.1371/journal.pgen.1011032&type=printable |
| work_keys_str_mv | AT ziyimo domainadaptiveneuralnetworksimprovesupervisedmachinelearningbasedonsimulatedpopulationgeneticdata AT adamsiepel domainadaptiveneuralnetworksimprovesupervisedmachinelearningbasedonsimulatedpopulationgeneticdata |