Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk

By Jian Zhou, Chandra L. Theesfeld, K. Yao, Kathleen M. Chen, Aaron K. Wong, Olga G. Troyanskaya. 2018

562 citations recorded by the source corpus.

Key challenges for human genetics, precision medicine and evolutionary biology include deciphering the regulatory code of gene expression and understanding the transcriptional effects of genome variation. However, this is extremely difficult because of the enormous scale of the noncoding mutation space. We developed a deep learning–based framework, ExPecto, that can accurately predict, ab initio from a DNA sequence, the tissue-specific transcriptional effects of mutations, including those that are rare or that have not been observed. We prioritized causal variants within disease- or trait-associated loci from all publicly available genome-wide association studies and experimentally validated predictions for four immune-related diseases. By exploiting the scalability of ExPecto, we characterized the regulatory mutation space for human RNA polymerase II–transcribed genes by in silico saturation mutagenesis and profiled > 140 million promoter-proximal mutations. This enables probing of evolutionary constraints on gene expression and ab initio prediction of mutation disease effects, making ExPecto an end-to-end computational framework for the in silico prediction of expression and disease risk. ExPecto is a deep learning–based framework that can predict the tissue-specific transcriptional effects of mutations on the basis of DNA sequence alone. ExPecto can prioritize causal variants from GWAS loci and be used to predict the disease risk of a variant.

Subfields: Genetics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Genetics

Citations in the corpus

CADD: predicting the deleteriousness of variants throughout the human genome (2018, positive): “Of special interest are regulatory variants in promoters, enhancers and near splice sites, as a number of other recent variant classifiers (26,52–55) have shown the potential of predicting regulatory effects from sequence and annotations describing the biological function.”
The Nucleotide Transformer: Building and Evaluating Robust Foundation Models for Human Genomics (2024, methodology): “Along this line of work, specialized deep learning (DL) models were trained to identify meaningful patterns of DNA, for example, to predict gene expression from DNA sequences alone [9, 10, 11, 12, 13], with recent work combining convolutional neural ∗Equal Supervision †Corresponding authors: t.”
Artificial Intelligence in Cancer Research and Precision Medicine (2021, background): “For example, AI may increasingly be used to help understand the functional impact of mutations, e.g., predict the impact of noncoding mutations on gene expression, epigenetic processes, as well as disease risk (51, 52).”
Obtaining genetics insights from deep learning via explainable artificial intelligence (2022, background): “Other common prediction targets include chromatin accessibility 11,12 , RNA binding 13 , gene expression 14–17 , splicing 18,19 or aspects of chromatin 3D organization 20 .”
Artificial intelligence in clinical and genomic diagnostics (2019, background): “Further extension to the ExPecto algorithm has demonstrated its ability to predict gene expression levels directly from DNA sequence information [70].”