EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations
Abstract Deep neural networks (DNNs) hold promise for functional genomics prediction, but their generalization capability may be limited by the amount of available data. To address this, we propose EvoAug, a suite of evolution-inspired augmentations that enhance the training of genomic DNNs by incre...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2023-05-01
|
| Series: | Genome Biology |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s13059-023-02941-w |
| _version_ | 1797832130427879424 |
|---|---|
| author | Nicholas Keone Lee Ziqi Tang Shushan Toneyan Peter K. Koo |
| author_facet | Nicholas Keone Lee Ziqi Tang Shushan Toneyan Peter K. Koo |
| author_sort | Nicholas Keone Lee |
| collection | DOAJ |
| description | Abstract Deep neural networks (DNNs) hold promise for functional genomics prediction, but their generalization capability may be limited by the amount of available data. To address this, we propose EvoAug, a suite of evolution-inspired augmentations that enhance the training of genomic DNNs by increasing genetic variation. Random transformation of DNA sequences can potentially alter their function in unknown ways, so we employ a fine-tuning procedure using the original non-transformed data to preserve functional integrity. Our results demonstrate that EvoAug substantially improves the generalization and interpretability of established DNNs across prominent regulatory genomics prediction tasks, offering a robust solution for genomic DNNs. |
| first_indexed | 2024-04-09T14:02:50Z |
| format | Article |
| id | doaj.art-fdf641cadc6b46828d142ef24c8d0c0d |
| institution | Directory Open Access Journal |
| issn | 1474-760X |
| language | English |
| last_indexed | 2024-04-09T14:02:50Z |
| publishDate | 2023-05-01 |
| publisher | BMC |
| record_format | Article |
| series | Genome Biology |
| spelling | doaj.art-fdf641cadc6b46828d142ef24c8d0c0d2023-05-07T11:14:46ZengBMCGenome Biology1474-760X2023-05-0124111410.1186/s13059-023-02941-wEvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentationsNicholas Keone Lee0Ziqi Tang1Shushan Toneyan2Peter K. Koo3Simons Center for Quantitative Biology, Cold Spring Harbor LaboratorySimons Center for Quantitative Biology, Cold Spring Harbor LaboratorySimons Center for Quantitative Biology, Cold Spring Harbor LaboratorySimons Center for Quantitative Biology, Cold Spring Harbor LaboratoryAbstract Deep neural networks (DNNs) hold promise for functional genomics prediction, but their generalization capability may be limited by the amount of available data. To address this, we propose EvoAug, a suite of evolution-inspired augmentations that enhance the training of genomic DNNs by increasing genetic variation. Random transformation of DNA sequences can potentially alter their function in unknown ways, so we employ a fine-tuning procedure using the original non-transformed data to preserve functional integrity. Our results demonstrate that EvoAug substantially improves the generalization and interpretability of established DNNs across prominent regulatory genomics prediction tasks, offering a robust solution for genomic DNNs.https://doi.org/10.1186/s13059-023-02941-wDeep learningRegulatory genomicsData augmentationsModel interpretability |
| spellingShingle | Nicholas Keone Lee Ziqi Tang Shushan Toneyan Peter K. Koo EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations Genome Biology Deep learning Regulatory genomics Data augmentations Model interpretability |
| title | EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations |
| title_full | EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations |
| title_fullStr | EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations |
| title_full_unstemmed | EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations |
| title_short | EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations |
| title_sort | evoaug improving generalization and interpretability of genomic deep neural networks with evolution inspired data augmentations |
| topic | Deep learning Regulatory genomics Data augmentations Model interpretability |
| url | https://doi.org/10.1186/s13059-023-02941-w |
| work_keys_str_mv | AT nicholaskeonelee evoaugimprovinggeneralizationandinterpretabilityofgenomicdeepneuralnetworkswithevolutioninspireddataaugmentations AT ziqitang evoaugimprovinggeneralizationandinterpretabilityofgenomicdeepneuralnetworkswithevolutioninspireddataaugmentations AT shushantoneyan evoaugimprovinggeneralizationandinterpretabilityofgenomicdeepneuralnetworkswithevolutioninspireddataaugmentations AT peterkkoo evoaugimprovinggeneralizationandinterpretabilityofgenomicdeepneuralnetworkswithevolutioninspireddataaugmentations |