< Back to Article

Genome Modeling System: A Knowledge Management Platform for Genomics

Fig 4

HCC1395 (“TST1”) example input, models, and outputs.

A test dataset for the HCC1395 cell line is provided with the GMS software to allow testing of software installation, and facilitate further development. It is also used to illustrate much of the current functionality of the GMS. HCC1395 tumor and the corresponding HCC1395BL ‘normal’ cell line DNA and RNA samples were sequenced by whole genome, exome, and RNA-seq methods producing six sets of instrument data for input to various GMS pipelines. Additional required inputs for the pipelines include a reference genome (e.g., GRCh37), gene annotations (e.g., Ensembl 67_37l), and variant databases (e.g., dbSNP37). Different versions (processing profiles) of the reference alignment were used to align WGS and exome DNA reads to the reference genome. A separate RNA-seq pipeline similarly aligns RNA reads. Alternate versions of the somatic variation pipeline are used to call various types of variants from exome and WGS data by comparing tumor and normal reference alignments. A differential expression pipeline identifies significantly altered transcript expression levels by comparing the tumor and normal RNA-seq alignments. Finally, the MedSeq pipeline summarizes all upstream pipelines into a single convenient result set. This includes a multitude of reports and visualizations for single nucleotide variants (SNVs), Indels (insertions and deletions), SVs (structural variants), CNVs (copy number variations), transcript fusions, differentially expressed genes, alternatively expressed isoforms, and much more. Data types are further integrated to, for example, identify which variants at the DNA level are expressed at the RNA level and which events affect known cancer driver genes or druggable targets.

Fig 4