Supporting data for "ChiRA: an integrated framework for Chimeric Read Analysis from RNA-RNA interactome and RNA structurome data"

With the advances in next-generation sequencing technologies, it is possible to determine RNA-RNA interaction and RNA structure predictions on a genome-wide level. The reads from these experiments usually are chimeric with each arm generated from one of the interaction partners. Due to short read lengths, often these sequenced arms ambiguously map to multiple locations. Thus, inferring the origin of these can be quite complicated. Here we present ChiRA, a generic framework for sensitive annotation of these chimeric reads, which in turn predict the sequenced hybrids. Grouping reference loci based on aligned common reads and quantification improved the handling of the multi-mapped reads in contrast to common strategies like the selection of the longest hit or a random choice among all hits. On benchmark data ChiRA improved the number of correct alignments to the reference up to 3-fold. It is shown that the genes that belong to the common read loci share the same protein families or similar pathways. In published data, ChiRA could detect 3 times as many new interactions compared to existing approaches. In addition, ChiRAViz can be used to visualize and filter large chimeric datasets intuitively.