Meta-clusterubf of PDAC cohorts using consensus clustering and COINCIDE approach.

Introduction

All the PDAC cohorts are clustered using meta genes and looked for similarity between clusters and generated meta-cluster using COINCIDE approach. Looked at functional aspect, pathways, gene signatures and biomarkers. Built single sample classifier, classifed PDXE, Pharmacogx, Organoids and Compass samples. Analyzed drugs specific to the subtypes.

Scripts

1.Coincide_implementation.R : The script is used for building COINCIDE framework for clustering all PDAC datasets. Survival analysis is also implemented in this script.

2.Meta_Effectsize_pathway_analysis: The script calculates the meta effect size and used for pathway analysis.

3.Gene_signature_score.R: Used the script to calculate gene signature score and compare across the metaclusters.

4.Known_biomarkers_scores.R: Used the script to compare known PDAC biomarkers across the metaclusters.

5.Single_Sample_Classifier_Metaclusters: Used for building the single sample classifier.

6.Subtyping_model_systems.R: Used single sample classifier for subtyping model systems like PDXE, celllines, COMPASS samples and Organoids.

7.CNA_Analysis.R: Compared the copy number aberration data between OUH and TCGA datasets

8.Cellularity_cohorts.R: Compared the cellularity of samples across cohorts and meta-clusters.

9.Compass_functionality_RECIST.R: Used to compare the difference between COMPASS metaclusters to find difference in tumor lesion, NLR and biomarkers.

10.Drug_Sensitivity_analysis: Performed drug sensitivity analysis for meta-clusters in celllines.

11.Centroid_based_clustering_Other_Classifiers.R: Clustered all the datasets using all published classification methods.

12.Combined_cohort_run.R: Script for combining celllines and PDAC datasets subtypes for combined heatmaps.

13.Common_between_published_classifeirs: Used to compare concordance between subtype calls in celllines cohorts like CCLE, GDSC1000, CTRPV2 and GCSI using different classifiers.