bioRxiv Science⌕ Search

Biology subjects

Selvarajan, S.

Publications and source records attributed to Selvarajan, S..

2 recordsLinked to original sources
bioRxiv2020

Phenotype Prediction using a Tensor Representation and Deep Learning from Data Independent Acquisition Mass Spectrometry

A novel approach for phenotype prediction is developed for mass spectrometric data. First, the data-independent acquisition (DIA) mass spectrometric data is converted into a novel file format called "DIA tensor" (DIAT) which contains all the peptide precursors and fragments information and can be used for convenient DIA visualization. The DIAT format is fed directly into a deep neural network to predict phenotypes without the need to identify peptides or proteins. We applied this strategy to a collection of 102 hepatocellular carcinoma samples and achieved an accuracy of 96.8% in classifying malignant from benign samples. We further applied refined model to 492 samples of thyroid nodules to predict thyroid cancer; and achieved a predictive accuracy of 91.7% in an independent cohort of 216 test samples. In conclusion, DIA tensor enables facile 2D visualization of DIA proteomics data as well as being a new approach for phenotype prediction directly from DIA-MS data.

bioinformatics↗
bioRxiv2020

DPHL: A pan-human protein mass spectrometry library for robust biomarker discovery using Data-Independent Acquisition and Parallel Reaction Monitoring

To answer the increasing need for detecting and validating protein biomarkers in clinical specimens, proteomic techniques are required that support the fast, reproducible and quantitative analysis of large clinical sample cohorts. Targeted mass spectrometry techniques, specifically SRM, PRM and the massively parallel SWATH/DIA technique have emerged as a powerful method for biomarker research. For optimal performance, they require prior knowledge about the fragment ion spectra of targeted peptides. In this report, we describe a mass spectrometric (MS) pipeline and spectral resource to support data-independent acquisition (DIA) and parallel reaction monitoring (PRM) based biomarker studies. To build the spectral resource we integrated common open-source MS computational tools to assemble an open source computational workflow based on Docker. It was then applied to generate a comprehensive DIA pan-human library (DPHL) from 1,096 data dependent acquisition (DDA) MS raw files, and it comprises 242,476 unique peptide sequences from 14,782 protein groups and 10,943 SwissProt-annotated proteins expressed in 16 types of cancer samples. In particular, tissue specimens from patients with prostate cancer, cervical cancer, colorectal cancer, hepatocellular carcinoma, gastric cancer, lung adenocarcinoma, squamous cell lung carcinoma, diseased thyroid, glioblastoma multiforme, sarcoma and diffuse large B-cell lymphoma (DLBCL), as well as plasma samples from a range of hematologic malignancies were collected from multiple clinics in China, the Netherlands and Singapore and included in the resource. This extensive spectral resource was then applied to a prostate cancer cohort of 17 patients, consisting of 8 patients with prostate cancer (PCa) and 9 with benign prostate hyperplasia (BPH), respectively. Data analysis of DIA data from these samples identified differential expressions of FASN, TPP1 and SPON2 in prostate tumors. Thereafter, PRM validation was applied to a larger PCa cohort of 57 patients and the differential expressions of FASN, TPP1 and SPON2 in prostate tumors were validated. As a second application, the DPHL spectral resource was applied to a patient cohort consisting of samples from 19 DLBCL patients and 18 healthy individuals. Differential expressions of CRP, CD44 and SAA1 between DLBCL cases and healthy controls were detected by DIA-MS and confirmed by PRM. These data demonstrate that the DPHL supported that DIA-PRM MS pipeline enables robust protein biomarker discoveries.

bioinformatics↗