tumordeconv/sim-data repository overview

⁠Decomprolute: Benchmarking study of proteomic based tumor deconvolution

A suite of scientific workflows to assess metrics to compare efficacy of protein-based tumor deconvolution algorithms. The goal of this project is to standardize the analysis and comparison of various tumor deconvolution datasets to compare their efficacy with different parameters.

⁠To Run

This repository contains all the tools needed to compare tumor deconvolution algorithms. So far we are focusing on comparing proteomic to mRNA measurements and assessing their correlation via the Spearman Rank statistic. To evaluate and see the results, you can:

cd perfEval
cwltool scatter-test.cwl scatter-test.yml

This will run the evaluation in our test YAML file. To update the parameters, create your own YAML file. The algorithm currently has five parameters:

1. mrna-algorithms: List of algorithms to use to deconvolve mRNA data. One of epic, xcell, cibersort, mcpcounter.
2. prot-algorithms: List of algorithms to use to deconvolve protein data. One of epic, xcell, cibersort, mcpcounter.
3. cancerTypes: List of cancer types
4. signatures: List of signature matrices, currently found in the signature matrix⁠ directory
5. tissueTypes: list of tissue types: tumor, normal, or all

⁠To add your algorithm

To test your own algorithm, please ensure that it will work with the signature matrices in this repository as well as the matrices generated in the mRNA and protein modules. Then deposit your CWL script and add your algorithm to the call-deconv-and-cor.cwl file.

⁠Example results

Currently the results produce two PDF files: one that plots the correlation across cell types for each tumor type/algorithm/tissue combination, and one that plots the correlation across patient cohorts.

⁠Architecture

We propose a modular architecture to enable 'plug and play' comparisons of different datasets and tools.The modules fall into three categories, each with a data collection and analysis module.

These modules are each describe below.

⁠Deconvolution algorithms

List here

⁠Deconvolution signatures

There are numerous ways to define the individual cell types we are using to run the deconvolution algorithms. We will upload specific lists to compare in our workflow.

⁠Data collection

We have collect pre-formatted sample data from the CPTAC Python API⁠ to better match the mRNA data. This CWL tool and Docker image are in the protData⁠ and mRNAdata⁠ directories.

Below are the available tumor types:

As such, datasets have been updated to following (added hnscc): ['brca', 'ccrcc', 'endometrial', 'colon', 'ovarian', 'hnscc', 'luad']