5744 - OncOS: scalable and accurate next-generation sequencing analytics for precision oncology and personalized patient care

Background

The advent of high-throughput next-generation sequencing (NGS) technologies has resulted in a deluge of data for a wide variety of clinical uses, with millions of samples sequenced to date. Data generated at an “-omics” level (genomics, transcriptomics, epigenomics, proteomics, etc) in cancer research and for clinical decision making is ushering in a new era of personalized cancer care. However this requires fast, accurate, and easily automatable bioinformatics pipelines capable of large scale analytics on big datasets without sacrificing accuracy. Here we present OncOS, a cloud-based auto-scaling architecture capable of performing highly accurate molecular profiling for personalized clinical insights.

Methods

A flexible cloud architecture implements bioinformatics pipelines dynamically depending on the input data, including a range of possible sample types (FFPE, fresh frozen, plasma/cfDNA), and clinical insights (clinical trial matching, drug matching, and genomic insights such as mutation calls, copy number variant calls, and MSI/TMB). This is powered by a pipeline scheduler and an elastic container service cluster that is capable of initializing a large number of elastic compute cloud instances for scalable and parallelized processing. Data is stored on HIPAA compliant and securely encrypted databases and simple storage services, with key information relayed to a web app for use in a clinical setting.

Results

OncOS has also been optimized for Positive Predictive Value (PPV), with testing on samples from the Multi-Center Mutation Calling in Multiple Cancers (MC3) project, a collaborative effort to provide a high confidence set of variants for patients in The Cancer Genome Atlas (TCGA). Benchmarking shows OncOS performs with a PPV of 87.4%, outperforming similar variant calling pipelines (BROAD institute 75.4%; MD Anderson 80.1%).

Conclusions

OncOS is a precision oncology platform with a cloud architecture capable of processing a variety of sample types at scale, optimized for variant calling PPV and drawing of key clinical insights.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Cambridge Cancer Genomics.

Disclosure

J.S. Thompson: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. J.H.R. Farmery: Shareholder / Stockholder / Stock options: Cambridge Cancer Genomics. H. Dobson: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. S. Frost: Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. J.W. Cassidy: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment, Officer / Board of Directors: Cambridge Cancer Genomics. N. Patel: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. H. Thompson: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics. H.W. Clifford: Leadership role, Shareholder / Stockholder / Stock options, Full / Part-time employment: Cambridge Cancer Genomics.