Date: 10/26/23
Speaker: Nilah Ioannidis
Title: TBA
Host: Manuel Rias
Personal genome interpretation: predicting transcriptome variation from sequence
Understanding the functional and clinical significance of personal genome variation is an important challenge in the field of precision medicine. A number of machine learning tools have been developed to predict the pathogenicity and molecular impact of individual variants of uncertain significance, but interpreting noncoding variation and determining the causal variants in trait-associated loci found in genome-wide association studies remains a challenge. I will discuss approaches for understanding the impact of personal genome variation on personal transcriptome variation, including methods developed for transcriptome-wide association studies and recent advances in genomic deep learning models that predict gene expression and other molecular phenotypes (such as chromatin accessibility, histone modifications, and transcription factor binding) from DNA sequence input. I will discuss the application of these methods to personal genome interpretation and some limitations of current genomic deep learning models when explaining variation in expression across individuals and predicting the direction of effect of cis-regulatory variation on expression.
Readings:
Preprint: https://www.biorxiv.org/content/10.1101/2023.06.30.547100v1
Review: https://www.nature.com/articles/s41576-019-0122-6