Immunopeptidome Cluster Profiling with Critical Variable Selection

Anonymous

Keywords: Unsupervised Machine Learning, Genomics, Immunology, Clustering

TL;DR: Immunopeptidome Cluster Profiling with Critical Variable Selection

Abstract: Despite the worldwide effort to gain a deeper comprehension of the molecular mechanism that characterizes cancer, its blueprint is still elusive. In the last decade, results in the context of cancer immunology helped researchers to identify a set of biological objects, namely antigens, which now represent a promising direction. Mass spectrometry is the typical investigation involved in collecting molecules. This method and modern techniques allow a limited but satisfactory throughout on peptide knowledge. However, the diversity coming out from such experiments is still poorly characterized. In this paper, UnChAnTies(Unsupervised characterization of antigen communities) is introduced. This tool is a computational platform that enables unsupervised clusterings and identifies critical variables characterizing each community of antigens. The goal of the pipeline is to identify relevant structures through unsupervised feature extraction with a model-free approach. Additionally, shed light, that takes part in the interplay between various kinds of cancer through their antigen are determined