作者: Umet Eser , L. Stirling Churchman
DOI: 10.1101/081380
关键词:
摘要: Numerous advances in sequencing technologies have revolutionized genomics through generating many types of genomic functional data. Statistical tools been developed to analyze individual data types, but there lack strategies integrate disparate datasets under a unified framework. Moreover, most analysis techniques heavily rely on feature selection and preprocessing which increase the difficulty addressing biological questions integration multiple datasets. Here, we introduce FIDDLE (Flexible Integration Data with Deep LEarning) an open source data-agnostic flexible integrative framework that learns representation from infer another type. As case study, use Saccharomyces cerevisiae predict global transcription start sites (TSS) simulation TSS-seq We demonstrate type can be inferred other sources without manually specifying relevant features preprocessing. show models built genome-wide perform profoundly better than Thus complex synergistic relationship within and, importantly, across