MSACL 2019 EU Abstract

High-Throughput PEAKS Workflow and SPIDER Algorithm for Large Scale Clinical Proteomics and Variant Identification Using PEAKS Online X

Jonathan R. Krieger, Julia Hopkins, Lei Xin, Baozhen Shan
Bioinformatics Solutions Inc.

Warning: Undefined variable $headshot in /var/www/html/view_abstract/view_abstract_in_program.php on line 704

Jonathan Krieger (Presenter) Bioinformatics Solutions Inc >> POSTER (PDF)

Presenter Bio: Jonathan Krieger completed his PhD at the University of Toronto in 2011. His PhD project used mass spectrometry to identify and quantify protein protein interactions associated with the Numb endocytic adaptor molecule. He then transitioned to a project manager in the SPARC Biocentre at SickKids, the institute's core proteomics facility where he managed the instrumentation as well as many of the projects and data analysis for the facility. He recently transitioned out of academia and has joined Bioinformatics Solutions Inc as both a senior applications scientist as well as the application and core service manager.

Relevant Financial Disclosures (within past 24 months) Salary BIoinformatics Solutions Inc. Conflict mitigation NOT REQUIRED.

NTRODUCTION: The field of clinical proteomics is rapidly advancing and new mass spectrometry technology as well as streamlined sample preparation workflows are making the use of mass spectrometry in the clinical lab ever more attractive.

METHODS: Large scale clinical projects, whether research or clinical, often require quantitation of thousands of proteins, either by label free (LFQ) or by using isobaric tags such as TMT. PEAKS Online is a new high-throughput protein sequencing software solution that runs on a shared resource, is flexible to scale, and is fully parallelized with the ability to run on any cluster or multi-cluster CPU machine. Herein we describe the use of PEAKS online in analyzing two published data sets; one employing LFQ, and the other employing TMT.

RESULTS: LFQ data with match between runs is notorious for taking a significant amount of time to search. With the new build of PEAKS Online X, we are able to search 28,858,408 MS2 Spectra (56 fractionated samples, 672 180-min Orbitrap MS Runs) in 17h11m with using 512 CPU Cores. For TMT data, we tested a CPTAC Study (Zhang et al, Nature, 2014) to test 155 colorectal cancer samples, each fractionated into 15 samples and run on an Orbitrap-Elite (FT-IT mode) for 180min/fraction. This resulted in 2325 total RAW Files accounting for 6975 total hours of MS acquisition time and 19,983,186 MS2 scans. In total, we identified 145,706 unique peptides and 11,321 protein groups compared to the initial published search of 124,823 and 7,526 peptides and proteins respectively. This was accomplished in under 48 hours of search time.
Finally, in clinical research the ability to identify sequences that are not necessarily in a standard database is critical to identifying disease causing mutations and patient specific differences. Our de novo based approach of generating de novo sequencing tags allows us to further investigate clinical datasets by identifying variants. Integrated seamlessly with PEAKS online is the SPIDER algorithm. SPIDER can identify these variants by matching de novo sequencing tags to database proteins by making allowances for homology peptide mutations. The SPIDER algorithm has the ability to detect differences between the de novo sequenced proteins and the sequences provided in a database, offering the power of identifying patient variants when a patient-specific database is not available.

CONCLUSIONS: PEAKS online provides an efficient, time-effective solution to search large-scale systems biology level proteomic experiments with unparalleled accuracy and the ability to identify patient specific variants in protein sequences.