CoFIM: A Computational Framework for Proteomic and Metabolomic Integrated Data Analysis

Motivation: Coronavirus disease (COVID-19) struck the world in late 2019 and caused millions of deaths worldwide as an infectious disease caused by the SARS-CoV-2 virus. An effective and early diagnosis is truly pivotal, and thus, many studies were initiated for that. The existing studies have some limitations such as only focusing on one type of omics data. The study aims to develop a computational model which studies COVID-19 with the integration of metabolomics and proteomics data, therefore reaching the goal of detecting the virus early in the stage. Methods: The computational framework for integrating multi-omics data (CoFIM) consists of two parts. The first part is a statistical analysis of datasets. In this study, a series of statistical analyses including univariate and multivariate analyses were conducted to identify a number of potential biomarkers after pulling the data of severe patients and non-severe patients from a proteomic and metabolomics dataset of sera samples of COVID-19 patients. The second part is a machine learning model that was conducted to predict a patient's disease progression and provide more insightful information to understand the disease. Results: CoFIM integrates both proteomic and metabolomics data and provides a customizable and scalable framework to analyze the multi-omics data. CoFIM is demonstrated on the COVID-19 dataset and a number of biomarkers were detected. Several new protein biomarkers (IGKV1-12, PCOLCE, PGLYRP2, PCYOX1, LUM, IGHV1-46) were detected. We believe CoFIM will be widely used for multi-omics data analysis.