This repository contains the workflow for tumour mass dormancy quantification across TCGA cancers.
This folder contains the code which establishes how the tumour mass dormancy and exhaustion programmes are predictive of enrichment of APOBEC mutational signatures.
- ApobecEnrich_define.R: Runs iterations of tSNE and expectation-maximization clustering to create Apobec mutational signature enrichment labels, and compare them to established Apobec enrichment scores.
- RandomForest_100sims.R: Creates 100 random forests from balanced APOBEC-enriched/non-enriched samples, evaluates the overall predictive capacity of the models, and details gene importance.
- RandomForest_exemplary.R: Creates a single random forest using the method as detailed above, and runs further visualisation, including association with normalised PLG expression (Fig 4F) and the corresponding multidimensional scaling plot (Fig 4G). Finally, runs Wilcoxon tests to determine differences in median expressions between Apobec-enriched/non-enriched samples.
- Zscore_normalization.R: Runs standard Z-score normalisation on exemplary FPKM-normalised data.
This folder contains code which is used to investigate links between tumour mass dormancy and specific cancer driver events.
- FishersExactTestAnalysis.R uses Fisher’s exact tests to identify known drivers of tumorigenesis with enrichment or depletion of mutations across all TCGA solid cancer primary tumour samples
- FishersExactTestAnalysis_cancerwise.R uses Fisher’s exact tests to identify known drivers of tumorigenesis with enrichment or depletion of mutations within individual TCGA cancer studies
- FishersExactTestAnalysis_hotspots.R uses Fisher’s exact test analysis to test for enrichment or depletion of HRAS, KRAS and NRAS recurrent hotspot mutations within TCGA samples exhibiting tumour mass dormancy
This folder contains the code for calculating hypoxia signature scores and associating these with tumour mass dormancy categories.
- Hypoxia_ScoreCalc.R: Calculates hypoxia scores using the signatures defined by Buffa, West, and Winter, following the procedure detailed by Bhandari et al. Compares the scores calculated here with those determined by Bhandari et al. from a slightly smaller TCGA cohort.
- Hypoxia_TMDCategories.R: Displays the varying distributions of Buffa signature hypoxia scores associated with the detailed tumour mass dormancy categories, as well as the shifting proportions of tumour mass dormancy categorisation associated with hypoxia scores at both the pan-cancer and cancer-specific levels.
- Hypoxia_DormancyCorrelation.R: Displays correlations between the hypoxia and various tumour mass dormancy-associated scores, as well as creating volcano plots visualising correlations on a cancer-specific basis.
Within this folder Immune_infiltration_analysis.R is used to calculate tumour microenvironment cell infiltration scores and their correlation with tumour mass dormancy.
Within this folder MutationalSignature_correlations.R is used to characterise the associations between tumour mass dormancy and various mutational signatures across the TCGA cohort.
Within this folder PHATE_analysis.R is used to perform PHATE dimensionality reduction on the TCGA solid cancer primary tumour samples based on their expression of genes within the tumour mass dormancy and exhaustion programmes.
Within this folder ProgrammeScoreCalculation.R is used to derive the per-sample dormancy, exhaustion and APOBEC programme scores, and ScoreDistributionAcrossCancers.R is used to visualise their density distributions.
Within this folder Programme_APOBEC_correlation.R is used to visualise the correlation between the mean expression of the AID/APOBEC gene family of enzymes and the dormancy and exhaustion programmes across the TCGA cohort.
Within this folder Intergene_correlations.R is used to visualise the correlations between the expression of individual members of the AID/APOBEC gene family and the genes belonging to the tumour mass dormancy and exhaustion programmes.
This folder contains the code for conducting survival analysis associated with tumour mass dormancy categorisation.
- Survival_TMD.R: Visualises survival curves using the pan-cancer cohort, and as a comparison of early/late stage cancers. Runs Cox proportional hazard models to account for features including age at diagnosis, gender, cancer type, and tumour stage in both binary (early/late) and full (Stage I-IV) modes. Conducts a Fisher's exact test to determine significant differences in tumour mass dormancy between early/late stage cancers.
Within this folder PatientMutationStatus.R is used to visualise the frequency of mutations making up the tumour mass dormancy, APOBEC and exhaustion programmes within the TCGA cohort, relative to the most frequently mutated COSMIC cancer drivers.
Within this folder TMD_classification.R is used to classify TCGA cohort samples into those displaying tumour mass dormancy.
Within this folder SelectionAnalysis.R is used to detect signals of positive selection with primary tumour samples with and without TMD using the dn/ds maximum-likelihood method.
This code is free and is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY. See the GNU General Public License for more details.