Moustafa Abdalla

This paper introduces peaBrain, a framework using convolutional neural networks to predict transcriptomic effects of non-coding variants, showing it outperforms linear models and identifies functional eQTLs.

The paper develops a framework to determine tissue-of-action for type 2 diabetes GWAS signals by integrating gene expression and epigenomic data, revealing key tissues and enhancing functional interpretation.

The paper analyzes de novo mutations in 1,465 genomes to show similar mutation rates across ancestries but reduced rates in the Amish population, with environmental factors possibly influencing mutation rates.