Correlate phenotypic traits against mutated genes, or against individual variants
This document describes genome music clinical-correlation version 0.04 (2013-05-14 at 16:03:04)
genome music clinical-correlation --bam-list=? --output-file=? [--maf-file=?] [--glm-clinical-data-file=?] [--use-maf-in-glm] [--skip-non-coding] [--skip-silent] [--clinical-correlation-matrix-file=?] [--input-clinical-correlation-matrix-file=?] [--genetic-data-type=?] [--numeric-clinical-data-file=?] [--numerical-data-test-method=?] [--categorical-clinical-data-file=?] [--glm-model-file=?]
... music clinical-correlation \
--bam-list /path/myBamList.tsv \
--maf-file /path/myMAF.tsv \
--numeric-clinical-data-file /path/myNumericData.tsv \
--genetic-data-type 'gene' \
--output-file /path/output_file
... music clinical-correlation \
--maf-file /path/myMAF.tsv \
--bam-list /path/myBamList.tsv \
--numeric-clinical-data-file /path/myNumericData.tsv \
--categorical-clinical-data-file /path/myClassData.tsv \
--genetic-data-type 'gene' \
--output-file /path/output_file
... music clinical-correlation \
--maf-file /path/myMAF.tsv \
--bam-list /path/myBamList.tsv \
--output-file /path/output_file \
--glm-model-file /path/model.tsv \
--glm-clinical-data-file /path/glm_clinical_data.tsv \
--use-maf-in-glm
Tab delimited list of \s-1BAM\s0 files [sample_name, normal_bam, tumor_bam] (See Description)
Results of clinical-correlation tool. Will have suffix added for data type
List of mutations using \s-1TCGA\s0 \s-1MAF\s0 specification v2.3
Clinical traits, mutational profiles, other mixed clinical data (See \s-1DESCRIPTION\s0)
Create a variant matrix from the \s-1MAF\s0 file as variant input to \s-1GLM\s0 analysis. Default value 'false' (--nouse-maf-in-glm) if not specified
Skip non-coding mutations from the provided \s-1MAF\s0 file Default value 'true' if not specified
Skip silent mutations from the provided \s-1MAF\s0 file Default value 'true' if not specified
Specify a file to store the sample-vs-gene matrix created during calculations
Instead of creating this from the \s-1MAF\s0, input the sample-vs-gene matrix for calculations
Correlate clinical data to \*(L"gene\*(R" or \*(L"variant\*(R" level data Default value 'gene' if not specified
Table of samples (y) vs. numeric clinical data category (x)
Either 'cor' for Pearson Correlation or 'wilcox' for the Wilcoxon Rank-Sum Test for numerical clinical data Default value 'cor' if not specified
Table of samples (y) vs. categorical clinical data category (x)
File outlining the type of model, response variable, covariants, etc. for the \s-1GLM\s0 analysis. (See \s-1DESCRIPTION\s0)
This command relates clinical traits and mutational data. Either one can perform correlation analysis between mutations recorded in a \s-1MAF\s0 and the particular phenotypic traits recorded in clinical data files for the same samples, or one can run a generalized linear model (\s-1GLM\s0) analysis on the same types of data.
The clinical data files for correlation must be separated between numeric and categoric data and must follow these conventions:
Headers are required
Each file must include at least 1 sample_id column and 1 attribute column, with the format being [sample_id clinical_data_attribute_1 clinical_data_attribute_2 ...]
The sample \s-1ID\s0 must match the sample \s-1ID\s0 listed in the \s-1MAF\s0 under \*(L"Tumor_Sample_Barcode\*(R" for relating the mutations of this sample.
Note the importance of the headers: the header for each clinical_data_attribute will appear in the output file to denote relationships with the mutation data from the \s-1MAF\s0.
Internally, the input data is fed into an R script which calculates a P-value representing the probability that the correlation seen between the mutations in each gene (or variant) and each phenotype trait are random. Lower P-values indicate lower randomness, or likely true correlations.
The results are saved to the output filename given with a suffix appended; \*(L".numeric.csv\*(R" will be appended for results derived from numeric clinical data, and \*(L".categorical.csv\*(R" will be appended for results derived from categorical clinical data. Also, \*(L".glm.csv\*(R" will be appended to the output filename for \s-1GLM\s0 results.
The \s-1GLM\s0 analysis accepts a mixed numeric and categoric clinical data file, input using the parameter --glm-clinical-data-file. \s-1GLM\s0 clinical data must adhere to the formats described above for the correlation clinical data files. \s-1GLM\s0 also requires the user to input a --glm-model-file. This file requires specific headers and defines the analysis to be performed rather exactly. Here are the conventions required for this file:
Columns must be ordered as such:
[ analysis_type clinical_data_trait_name variant/gene_name covariates memo ]
The 'analysis_type' column must contain either \*(L"Q\*(R", indicating a quantative trait, or \*(L"B\*(R", indicating a binary trait will be examined.
The 'clinical_data_trait_name' is the name of a clinical data trait defined by being a header in the --glm-clinical-data-file.
The 'variant/gene_name' can either be the name of one or more columns from the --glm-clinical-data-file, or the name of one or more mutated gene names from the \s-1MAF\s0, separated by \*(L"|\*(R". If this column is left blank, or instead contains \*(L"\s-1NA\s0\*(R", then each column from either the variant mutation matrix (--use-maf-in-glm) or alternatively the --glm-clinical-data-file is used consecutively as the variant column in independent analyses.
'covariates' are the names of one or more columns from the --glm-clinical-data-file, separated by \*(L"+\*(R".
'memo' is any note deemed useful to the user. It will be printed in the output data file for reference.
\s-1GLM\s0 analysis may be performed using solely the data input into --glm-clinical-data-file, as described above, or alternatively, mutational data from the \s-1MAF\s0 may be included as variants in the \s-1GLM\s0 analysis, as also described above. Use the --use-maf-in-glm flag to include the mutation matrix derived from the maf as variant data.
Note that all input files for both correlation and \s-1GLM\s0 analysis must be tab-separated.
Copyright (C) 2010-2011 Washington University in St. Louis.
It is released under the Lesser \s-1GNU\s0 Public License (\s-1LGPL\s0) version 3. See the associated \s-1LICENSE\s0 file in this distribution.
Nathan D. Dees, Ph.D. Qunyuan Zhang, Ph.D. William Schierding, M.S.
genome-music(1), genome(1)