Endoseq Project

Genetic ancestries and their relationship with endometrial cancer outcomes

Endoseq ancestry analyses explore the relationship of Local ancestry inference with two outcomes observed in endometrial cancer (i.e. Somatic mutations counts and histopathological characteristics). Local ancestry was performed using RFMIX ver 2 with four super population groups as references. The analyses performed were:

1. Comparing Genome-wide average rate of somatic mutations across diplotypes in each individual:
   The script Fig3C_Comparing_Genome-wide_Average_of_Sommuts_Across_Diplotypes.R determine (i) the total lenght for each diplotype (i.e. Eur/Eur, Afr/Afr, and Eur/Afr), (ii) the number of somatic mutations per diplotype per individual, and using both results will determine the genome-wide rate of somatic mutations per diplotype. After determine the average, the script will estimate the pairwise differences among diplotypes per individual and will plot the distribution of these differences. We determine if the difference is significantly away from zero as an indication that one average is greater than other averages.

2. Firth Regression to explore the relationship of local ancestry with total mutation burden (TMB) and histopathological outcomes:
   The script AdmixtureMapping_FirthRegression_HistopathologicalOutcomes.R generates a table of haplotype counts per locus for each individual from the local ancestry data (0,1, and 2 African counts in an specific loci). With this information coupled with clinical data (e.g. Age of Diagnosis, histopathological type, etc) and principal components, this script will run a Firth regression and a manhattan plot with significant values annotated.

3. Rare variant accumulation and its relationship with the histopathological outcome:
   Germline variants were annotated using SnpEff and filtered based on its functional characteristics (missense and nonsense) using SnpSift. Rare variant analysis was perfomed using the script SKAT_rare_variant_analysis_Endoseq.R with the SKAT packages and then filtered for genes that were included in the genomic region with the most significant signal of the First regressions.