QproMS: a web application for label-free proteomic data analysis

Abstract Motivation Proteomics has experienced a growth in methods and data analysis approaches, with the development of new data-dependent (DDA) and data-independent acquisition (DIA) approaches and several search engine algorithms and software packages. Each of these workflows has its unique data analysis package that performs data reduction, missing value imputation, statistical testing and visualisation. Often, these tools are dedicated to an expert user. Results We present Quantitative Proteomics Made Simple (QProMS), a user-friendly, search engine-agnostic data analysis and visualisation pipeline. QProMS guides the user through data analysis and statistical testing in a graphical interface. Statistical tests rely on established R functions and are compatible with all types of label-free quantification experiments. The pipeline recapitulates features from different available software packages and introduces mixed imputation, an improved framework for handling missing values that does not rely on machine learning. QProMS can also perform interaction analyses based on gene ontology, or by querying protein-protein interaction databases. All figures in QProMS are interactive, allowing for investigation of individual proteins of interest before exporting. The analysis can be saved in a standalone report. QProMS provides a platform for reproducible proteomic data analysis for novice and experienced users, enabling state-of-the art data analysis of a wide variety of label-free proteomic workflows ranging from global proteome profiling to targeted methods such as proximity labelling. Availability and implementation QproMS is accessible as a web server hosted at https://shiny.bioserver.ieo.it/app/qproms or can be run locally as a standalone shiny application with the code and instructions provided at https://github.com/ieoresearch/QProMS. The application may also be run locally by installing as a library and running a single command as described in the README. Code to generate benchmarking is available at https://github.com/grandrea/mixed-imputation-benchmark.