Synthesizing genome regulation data with vote-counting. Issue 12 (December 2022)

• Record Type: Journal Article
• Authors: Fischer, Martin; Hoffmann, Steve
• Is Part Of: Trends in genetics. Volume 38:Issue 12(2022)
• Page Start: 1208
• Page End: 1216
• Publication Date: 2022-12
• Subjects: meta-analysis -- transcriptomics -- genomics -- p53 -- epigenome -- enhancers; Genetics -- Periodicals; 576.5
• DOI: 10.1016/j.tig.2022.06.012
• Languages: English
• ISSNs: 0168-9525

Abstract

Abstract : The increasing availability of high-throughput datasets allows amalgamating research information across a large body of genome regulation studies. Given the recent success of meta-analyses on transcriptional regulators, epigenetic marks, and enhancer:gene associations, we expect that such surveys will continue to provide novel and reproducible insights. However, meta-analyses are severely hampered by the diversity of available data, concurring protocols, an eclectic amount of bioinformatics tools, and myriads of conceivable parameter combinations. Such factors can easily bar life scientists from synthesizing omics data and substantially curb their interpretability. Despite statistical challenges of the method, we would like to emphasize the advantages of joining data from different sources through vote-counting and showcase examples that achieve a simple but highly intuitive data integration. Highlights: An avalanche of omics data generated by multiple high-throughput experiments provides an opportunity to validate results and to make new discoveries. While typically regarded as a measure of last resort for meta-analyses, naive vote-counting approaches can have considerable advantages when synthesizing genome regulation data and their simplicity enables a more significant number of researchers to use meta-analyses. Powerful genomic signatures can be extracted via meta-analyses. Vote-counting-based procedures enable the user to score and rank the results to intuitively balance the tradeoff between specificity and sensitivity. Vote-counting-based scores and ranks allow for a swift integration of diverse data types (e.g., ChIP-seq-derived transcription factor binding data and RNA-seq-derived differential gene expression data).