Updates to a Lab-Developed Bioinformatics Pipeline and Application for Interpretation of Clinical Next-Generation Sequencing Panels

Abstract

Abstract Objectives Our goal was to enhance our next-generation sequencing (NGS) molecular oncology workflow from sequencing to analysis through improvements to our custom-built and previously described NGS application. Methods Over 1 year, we collected feedback regarding workflow pain-points and feature requests from all end users of our NGS application. The application consists of a series of scripted pipelines, a MySQL database, and a Java Graphic User Interface (GUI); the end users include molecular pathologists (MPs), medical technologist/medical laboratory technologists (MTs/MLTs), and the molecular laboratory manager. These feedback data were used to engineer significant changes to the pipelines and software architecture. These architecture changes provided the backbone to a suite of feature enhancements aimed to improve turnaround time, decrease manual processes, and increase efficiency for the molecular laboratory staff and directors. Summary The key software architecture changes include implementing support for multiple environments, refactoring common code in the different pipelines, migrating from a per-run pipeline model to a per-sample pipeline model, and key updates to the MySQL database. These changes enabled development of many technical and user experience improvements. We eliminated the need for the pipelines to be launched manually from the Linux command line. Multiple pipelines can be executed concurrently. We created a per-sample pipeline status monitor. Sample entry is integrated with our Laboratory Information System (LIS) barcodes, thus reducing the possibility of transcription errors. We developed quality assurance reports. Socket-based integration with Integrated Genomics Viewer (IGV) was enhanced. We enabled rapid loading of key alignment data into IGV over a wireless network. Features to support resident and fellow driven variant and gene annotation reporting were developed. Support for additional clinical databases was implemented. Conclusions The designed feature enhancements to our previously reported NGS application have added significant sophistication and safety to our clinical NGS workflow. For example, our NGS consensus conference can be held in a conference room over a wireless network, and a trainee can prepare and present each case without ever leaving the application. To date, we have analyzed 2,540 samples using three different assays (TruSight Myeloid Sequencing Panel, AmpliSeq Cancer Hotspot Panel, GlioSeq) and four sequencing instruments (NextSeq, MiSeq, Proton, PGM) in this application. The code is freely available on GitHub.