Scripts for reproducing "Lung cancer subtyping using cell-free DNA fragmentomes and protein biomarkers".
Does PCA on CLCGP 1mb CN signals and trains models to predict subtype in CLCGP. Predictions from LEMA subtype are generated by applying these models to PCs of the LEMA CN signals projected CLCGP subspace.
Cross-validated results for LUAD vs LUSC in LEMA on the subset of LEMA with proteins measured. Multimodal model includes 5 protein measurements together with the top 8 PCs from the projection onto CLCGP.