DNA language models are powerful predictors of genome-wide variant effects
Языковые модели ДНК являются мощными предикторами эффектов вариантов по всему геному
2023-10-26
SCID: 54.1/fhkd236p
Discuss with AI
Arabidopsis thalianaDNA language modelsGenomic Pre-trained Networkgenome-wide variant effectsunsupervised pretraining
Figures from the paper
Abstract (AI)
Abstract The expanding catalog of genome-wide association studies (GWAS) provides biological insights across a variety of species, but identifying the causal variants behind these associations remains a significant challenge. Experimental validation is both labor-intensive and costly, highlighting the need for accurate, scalable computational methods to predict the effects of genetic variants across the entire genome. Inspired by recent progress in natural language processing, unsupervised pre-training on large protein sequence databases has proven successful in extracting complex information related to proteins. These models showcase their ability to learn variant effects in coding regions using an unsupervised approach. Expanding on this idea, we here introduce the G enomic P re-trained N etwork ( GPN ), a model designed to learn genome-wide variant effects through unsupervised pre-training on genomic DNA sequences. Our model also successfully learns gene structure and DNA motifs without any supervision. To demonstrate its utility, we train GPN on unaligned reference genomes of Arabidopsis thaliana and seven related species within the Brassicales order, and evaluate its ability to predict the functional impact of genetic variants in Arabidopsis thaliana by utilizing allele frequencies from the 1001 Genomes Project and a comprehensive database of GWAS. Notably, GPN outperforms predictors based on popular conservation scores such as phyloP and phastCons. Our predictions for Arabidopsis thaliana can be visualized as sequence logos in the UCSC Genome Browser ( https://genome.ucsc.edu/s/gbenegas/gpn-arabidopsis ). We provide code ( https://github.com/songlab-cal/gpn ) to train GPN for any given species using its DNA sequence alone, enabling unsupervised prediction of variant effects across the entire genome.
Key Findings
1
GPN learns biologically meaningful gene structures and DNA motifs without supervised training.
2
GPN outperforms widely used conservation-based predictors, including phyloP and phastCons, in evaluating variant functional effects.
3
The Genomic Pre-trained Network (GPN) predicts genome-wide genetic variant effects by unsupervised pretraining on genomic DNA sequences.
4
The released code enables species-specific, sequence-only training and unsupervised genome-wide variant-effect prediction.
5
Trained on unaligned genomes from Arabidopsis thaliana and seven related Brassicales species, GPN predicts variant impacts in A. thaliana.
Research Object
genome-wide genetic variants in Arabidopsis thaliana and related Brassicales species
Research Subject
functional effects and predictive signatures of genetic variants across the genome, including gene structure and DNA motifs
Publication Details
Publication Date
2023-10-26
Journal
Publisher
ISSN
Cited by
162
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai10
Scikit-learn: Machine Learning in Python2012
Highly accurate protein structure prediction with AlphaFold2021
The Human Genome Browser at UCSC2002
HuggingFace's Transformers: State-of-the-art Natural Language Processing2019
DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome2021
Scaling Laws for Neural Language Models2020
Sparks of Artificial General Intelligence: Early experiments with GPT-42023
Ten things you should know about transposable elements2018
A guide for the diagnosis of rare and undiagnosed disease: beyond the exome2022
Qualitative Spatial Reasoning over Questions (Short Paper)2022
Cited by4
Species-aware DNA language models capture regulatory elements and their evolution2024
Cross-species modeling of plant genomes at single-nucleotide resolution using a pretrained DNA language model2025
BEND: Benchmarking DNA Language Models on biologically meaningful tasks2023
Advancing bioinformatics with language models: components, applications, and perspectives2026