Abstract

MotivationSmall insertion and deletion (sindel) of human genome has an important implication for human disease. One important mechanism for noncoding sindel to have an impact on human diseases and phenotypes is through the regulation of gene expression. Nevertheless, current sequencing technology may lack statistical power and resolution to pinpoint the causal sindel due to lower minor allele frequency or small effect. As an alternative solution, a supervised machine learning method can identify the otherwise missing causal sindels by predicting the regulatory potential of sindels directly. However, computational methods for annotating and predicting the regulatory sindels, especially in the noncoding regions, are underdeveloped. ResultsBy leveraging recognized sindels in cis-expression quantitative trait loci (cis-eQTLs) across 44 tissues and cell types in GTEx, and a compilation of both generic functional annotations and tissue/cell typespecific multi-omics features generated by a sequence-based deep learning model, we developed TIVAN-indel, which is an XGBoost-based supervised framework for scoring noncoding sindels based their potential to regulate the nearby gene expression. As a result, we demonstrate that TIVAN-indel achieves the best prediction performance in both cross-validation with-tissue prediction and independent cross-tissue evaluation. As an independent evaluation, we train TIVAN-indel from "Whole Blood" tissue in GTEx data and test the model using 15 immune cell types from an independent study DICE. Lastly, we perform an enrichment analysis for both recognized and predicted sindels in key regulatory regions such as chromatin interactions, open chromatin and histone modification sites, and find biologically meaningful enrichment patterns. Availability and implementationhttps://github.com/lichen-lab/TIVAN-indel Contactli.chen1@ufl.edu