README.md

MDS_MOFA

Scripts and data used in the Multi modal MDS analysis paper

How to use the scripts

matrices_preparation_for_MOFA: Create list of matrices of same samples' different modalities for both cohort separately
BMMNC_MOFA_04012023.Rmd: Downstream analysis of latent factors from MOFA, including survival analysis for BMMNC cohort
CD34RNAseq_MOFA_04012023.Rmd: Downstream analysis of latent factors from MOFA, including survival analysis for CD34+ RNASeq cohort
Characterization_of_mutations_relation_with_geneSets.R: Characterization of SF3B1 & SRSF2 mutation association with gene sets scores, including survival analysis

Example figure generation

Heatmap of important features within views per factors (Fig 2 a-b)

Run the "Important features per factors" chunk in BMMNC_MOFA_04012023.Rmd. Important features were selected by having absolute weights over 0.5.

Heatmap of important features within Factor 1 (Fig 2 c-d)

Run the "Important features for Factor 1" chunk in BMMNC_MOFA_04012023.Rmd. Important features were selected by having absolute scaled weights over 0.5 for Factor 1.

Correlogram of mutations and gene sets (Fig 5a & 6a)

The "corrTabs" function takes gene sets scores and mutations and calcutes their correlation coeffecients in addition to p-values from Wilcox rank-sum test (Characterization_of_mutations_relation_with_geneSets.R). Heatmap then generated for each mutation separately. Stars indicating p-values manually added.

Boxplots of significant gene sets and mutation status (Fig 5b-d & 6b-d)

The "my_Plot_class" function use significant gene sets names as an argument and create boxplots for each of them with mutation status. For each mutation function run separately (Characterization_of_mutations_relation_with_geneSets.R).

Kaplan-Meier plots for gene set based on mutation status

The "BMMNC/CD34+ Clinical Outcomes for Aging signatures of patients based on SF3B1 status" parts in the Characterization_of_mutations_relation_with_geneSets.R create Kaplan-Meier plots for specific Inflammation/Aging gene sets (Inflammator chemokines and Inflammatory cytokines, first splitting them based on SF3B1 status (WT and mutated) then splitting mutated ones into "high" and "low". Pairwise p-values and legend titles manually added.