17. Djerba: A modular system to generate clinical genome interpretation reports for cancer.

We present Djerba, a software package for the translation of bioinformatic pipeline output from individual tumor genomes and transcriptomes into clinical reports for precision cancer medicine in a CAP/CLIA/ACD accredited laboratory. Specific use cases are clinical genome and transcriptome sequencing for therapeutic assignment; along with cell-free DNA sequencing using targeted panels and whole genome sequencing, for early cancer and minimal residual disease detection. The modular structure of Djerba enables it to process a wide variety of data, with reports customized for available inputs; rapid development of new reporting modes; and openness to collaboration between multiple researchers and institutions. Clinical reporting data derives from assays such as whole genome tumour/normal and whole transcriptome sequencing. Djerba inputs results from one or more bioinformatics pipelines, with tools including but not limited to GATK, VariantEffectPredictor, Sequenza, Delly, RSEM, Star-Fusion, Arriba, and MAVIS. Djerba then queries external resources such as OncoKB, for relevant clinical annotation and any recommended treatments; generates summary statistics, plots and tables with links to further information; receives a brief text summary written by a cancer genome interpretation specialist; collates its results into a machine-readable JSON document; and finally renders its output as a PDF report for use by clinicians. We upload the JSON report documents to a CouchDB database, to facilitate comparison and tracking across multiple patients. Djerba enables fast, flexible collation of genomic analysis and variant calling into a standardized, compact reporting document, in both human-readable and machine-readable formats. [ABSTRACT FROM AUTHOR]