Bioinformatic pipelines for whole transcriptome sequencing data exploitation in leukemia patients with complex structural variants

Background Extensive genome rearrangements, known as chromothripsis, have been recently identified in several cancer types. Chromothripsis leads to complex structural variants (cSVs) causing aberrant gene expression and the formation of de novo fusion genes, which can trigger cancer development, or worsen its clinical course. The functional impact of cSVs can be studied at the RNA level using whole transcriptome sequencing (total RNA-Seq). It represents a powerful tool for discovering, profiling, and quantifying changes of gene expression in the overall genomic context. However, bioinformatic analysis of transcriptomic data, especially in cases with cSVs, is a complex and challenging task, and the development of proper bioinformatic tools for transcriptome studies is necessary. Methods We designed a bioinformatic workflow for the analysis of total RNA-Seq data consisting of two separate parts (pipelines): The first pipeline incorporates a statistical solution for differential gene expression analysis in a biologically heterogeneous sample set. We utilized results from transcriptomic arrays which were carried out in parallel to increase the precision of the analysis. The second pipeline is used for the identification of de novo fusion genes. Special attention was given to the filtering of false positives (FPs), which was achieved through consensus fusion calling with several fusion gene callers. We applied the workflow to the data obtained from ten patients with chronic lymphocytic leukemia (CLL) to describe the consequences of their cSVs in detail. The fusion genes identified by our pipeline were correlated with genomic break-points detected by genomic arrays. Results We set up a novel solution for differential gene expression analysis of individual samples and de novo fusion gene detection from total RNA-Seq data. The results of the differential gene expression analysis were concordant with results obtained by transcriptomic arrays, which demonstrates the analytical capabilities of our method. We also showed that the consensus fusion gene detection approach was able to identify true positives (TPs) efficiently. Detected coordinates of fusion gene junctions were in concordance with genomic breakpoints assessed using genomic arrays. Discussion Byapplying our methods to real clinical samples, we proved that our approach for total RNA-Seq data analysis generates results consistent with other genomic analytical techniques. The data obtained by our analyses provided clues for the study of the biological consequences of cSVs with far-reaching implications for clinical outcome and management of cancer patients. The bioinformatic workflow is also widely applicable for addressing other research questions in different contexts, for which transcriptomic data are generated.