PT-005 - GRIN2 IS A POWERFUL TOOL TO IDENTIFY GENOMIC HOT-SPOT LOCI AND ENABLES MULTI-OMICS INTEGRATION.

A. Elsayed¹, C. Mullighan¹, S. Pounds²; ¹St. Jude Children's Research Hospital, Memphis, TN, USA, ²Member, St. Jude Faculty Director of Biostatistics Courses, St. Jude Graduate School of Biomedical, Memphis, TN, USA.

Background: Genomic lesions are considered among the primary drivers of oncogenesis for many cancers. Recent advances in sequencing technologies have provided us with an exciting opportunity to identify multiple types of genomic lesions such as single-nucleotide substitutions, insertions, deletions, and structural rearrangements. There is a great need to analyze and integrate this information to identify important targets in cancer.

Methods: We first introduced the Genomic Random Interval (GRIN) method in 2013 to identify genomic loci with a statistically significant overabundance of a specific type of lesion. Since then, we have improved the GRIN method so that it can identify hot spot loci that are affected by a constellation of multiple types of lesions (SNVs, gain, loss, etc..), while other available tools such as GISTIC2.0 and MutSigCV can only evaluate one type of lesions. We also made some technical improvements that greatly increased computing speed and added convenience features to import annotation data for genes and regulatory features automatically from reputable public bioinformatic databases. We have also incorporated the ability to evaluate the association of lesions with transcriptome data at the individual transcript and pathway levels, thereby adding additional layers of evidence to identify candidate driver events more confidently.

Results: Statistical simulation studies and an example analysis of a dataset that encompasses 466 children or young adults newly diagnosed with T-cell Acute Lymphoblastic Leukemia (T-ALL) enrolled on multiple St. Jude and Children’s Oncology Group protocols identified multiple novel coding and non-coding hot-spot loci not captured by other tools. In addition, association analysis between lesion and expression data on the pathway level identified small groups of patients with distinct expression profiles that were found associated with poor treatment outcomes who might benefit from targeted treatment strategies.

Conclusion: We introduce GRIN2.0 tool that can identify genomic hot-spot loci affected by one or multiple types of genomic lesions and will offer the research community an exciting opportunity for multi-omics integration which can lead to biological discoveries that are invisible to other methods.