The associated publication reports proteogenomic analysis of non-small cell lung cancer (NSCLC), where we identified molecular subtypes with distinct immune evasion mechanisms and therapeutic targets, and validated our classification method in separate clinical cohorts. This protocol describes sections of the bioinformatics analysis of the multi-omics data, namely, data analysis and processing for panel sequencing, identification of cancer- and driver-related proteins in proteomics data, proteogenomics search, and machine learning-based classifiers for NSCLC subtyping. Specifically, a cohort classifier was built using support-vector machine-recursive feature elimination (SVM-RFE) algorithm applied to in-depth proteomics data from a cohort of 141 samples. The classifier was then validated in three external datasets. Another classifier, suitable for single-sample subtyping, was built using k-top scoring pairs (k-TSP) algorithm applied to label-free data from a cohort of 136 samples. The k-TSP-based classifier was validated in two independent cohorts and an additional external dataset.