Pre-cancer colorectal polyps: HTAN Vanderbilt dataset. Chen et al. Cell 2021 (Discovery & Validation Sets)

[rmadupuri]

Fixes #1915
Testing Instance:
Triage Portal: https://triage.cbioportal.mskcc.org/study/summary?id=crc_hta11_htan_2021
Private Portal: https://private.cbioportal.mskcc.org/study/summary?id=crc_hta11_htan_2021

Curation and transformation of Pre-cancer HTAN CRC Vanderbilt Dataset:

Data collection:

• Data Source: Data on HTAN Portal
• Reference publication: PubMed Reference
• Reference genome used: GRCh37 for WXS and GRCh38 for scRNA-seq
• Datatypes available: Clinical, WES MAF, scRNA (transformed to absolute and relative cell counts, as well as pseudo-bulk RNA counts), MxIF, and HE images.

Sample size selection

• Discovery & Validation set samples were used to generate the study (61 polyps). 30 samples were whole exome sequenced. 55 samples went though scRNA-seq exp analysis.
• Two patients (HTA11_3918 and HTA11_4781) were not listed in the supplementary clinical tables but were included in the Oncoplot analysis (Figure 1). They have been added to the cohort to match the oncoplot.

Clinical data

• Patient-Level Data: Table S1, Participants tab in HTAN Portal
• Sample-Level Data:Table S1, Biospecimens tab in HTAN Portal

Mutation data

• The Level 3 Bulk DNA filtered data files were obtained from the Vanderbilt team.
• Variants were annotated using Genome Nexus.

scRNA-seq data

• The Discovery, Validation set h5ad files for both epithelial and non-epithelial cells were used from cellxgene
  • Files used: Discovery (DIS) set of human colorectal tumor: Epithelial && Validation (Val) set of human colorectal tumor: Epithelial && VAL and DIS datasets: Non-Epithelial. They contain the normalized counts.
• The absolute and relative cell frequencies in Generic Assay format were calculated
• The pseudo bulk RNA expression counts per sample was calculated from scRNA-seq by averaging the values across the cells.
• The script to generate the absolute, relative cell freq and pseudo bulk RNA-seq data from h5ad files : https://gist.github.com/rmadupuri/447b5689b256ccc4880aa26fb91b72ff
• The pseudo bulk RNA expression data was log transformed and zscores were calculated.

Imaging data

• H&E data was available for 26 samples: H&E tab
• MxIF images were available for 25 samples: MxIF tab. Multiple images were available per sample and as we do not support this, the images were split to multiple tabs for now as MxIF Image 1, 2, 3..