Global ancestry estimates for the TCGA panCancer cohort from ADMIXTURE
The Cancer Genome Atlas (TCGA) is a vital resource in molecular cancer research. Opportunities to conduct cancer health disparities research from this resource are currently limited by incomplete data capture for self-reported race. Moreover, self-reported measures have known limitations, such as binning mixed race individuals into a single racial group which may not reflect their genetic make-up and thus risk. Therefore, we estimated global ancestry for all available TCGA samples according to standardized populations from 1000 Genomes.
For all available sample types (primary solid tumor, blood derived normal or other), genotypes were downloaded from TCGA’s Legacy Archive. In total there were 22,963 samples from 11,127 TCGA participants over 30 cancers included.
ADMIXTURE software was used to estimate ancestral proportions from each of the five 1000 Genomes global super populations. Phase 3 samples from 1000 Genomes (n = 2504) were used as reference.
Super populations:
- African (AFR)
- Admixed American (AMR)
- East Asian (EAS)
- European (EUR)
- South Asian (SAS)
- admixture_calls.txt
- ID - TCGA ID
- POP - dominant super population
- EUR:AFR - ADMIXTURE global ancestry estimates for 5 super populations
- tissue - tissue type
- admixture_calls_se.txt
- ID - TCGA ID
- EUR:AFR - standard errors from 200 boostrapped replicates
- tissue - tissue type
- admixture_calls_by_chr.txt and admixture_calls_se_by_chr.txt
- Contain same information as admixture_calls.txt and admixture_calls_se.txt but also include chromosome for each set of results
- entropy.txt
- ID - TCGA ID, entropy - Shannon's entropy, tissue - tissue type
- supervised_snp_list.txt
- Approximately 700,000 variants that overlapped between TCGA and 1000 Genomes used for ancestry estimation
- X1 - chromosome, X2 - SNP name, X3 - Position, X4 - base-pair coordinate, X5 - allele 1 (usually minor), X6 - allele 2 (usually major)
- blood_derived_normal_pca.txt, primary_solid_tumor_pca.txt, and other_tissues_pca.txt
- First 20 PCs by tissue type (analysis performed in plink)
- ID - TCGA ID, tissue - tissue PCA performed in, PC1:PC20 - first 20 PCs in order
- Data are also available at OSF
- stepByStep
- contains step by step instructions for downloading/cleaning files and running ADMIXTURE
- stepByStepSupervised
- incorporating 1000 Genomes data and performing the supervised analysis
Jordan Creed
Travis Gerke
Any questions or comments concerning the data or processes described in this repo can be directed to Jordan Creed @ Jordan.H.Creed@moffitt.org or Travis Gerke @ Travis.Gerke@moffitt.org.