Churchman, L. S. | Search bioRxiv

bioRxiv2018

Spt6 is required for the fidelity of promoter selection

Spt6 is a conserved factor that controls transcription and chromatin structure across the genome. Although Spt6 is viewed as an elongation factor, spt6 mutations in Saccharomyces cerevisiae allow elevated levels of transcripts from within coding regions, suggesting that Spt6 also controls initiation. To address the requirements for Spt6 in transcription and chromatin structure, we have combined four genome-wide approaches. Our results demonstrate that Spt6 represses transcription initiation at thousands of intragenic promoters. We characterize these intragenic promoters, and find sequence features conserved with genic promoters. Finally, we show that Spt6 also regulates transcription initiation at most genic promoters and propose a model of initiation-site competition to account for this. Together, our results demonstrate that Spt6 controls the fidelity of transcription initiation throughout the genome and reveal the magnitude of the potential for expressing alternative genetic information via intragenic promoters.

genomics↗

bioRxiv2016

FIDDLE: An integrative deep learning framework for functional genomic data inference

Numerous advances in sequencing technologies have revolutionized genomics through generating many types of genomic functional data. Statistical tools have been developed to analyze individual data types, but there lack strategies to integrate disparate datasets under a unified framework. Moreover, most analysis techniques heavily rely on feature selection and data preprocessing which increase the difficulty of addressing biological questions through the integration of multiple datasets. Here, we introduce FIDDLE (Flexible Integration of Data with Deep LEarning) an open source data-agnostic flexible integrative framework that learns a unified representation from multiple data types to infer another data type. As a case study, we use multiple Saccharomyces cerevisiae genomic datasets to predict global transcription start sites (TSS) through the simulation of TSS-seq data. We demonstrate that a type of data can be inferred from other sources of data types without manually specifying the relevant features and preprocessing. We show that models built from multiple genome-wide datasets perform profoundly better than models built from individual datasets. Thus FIDDLE learns the complex synergistic relationship within individual datasets and, importantly, across datasets.

bioinformatics↗