Differentially Expressed GTEx V8 Tissues

Differential Gene Expression Analysis

This notebook generates the sex-biased differential gene expression analysis. Differential Analysis (DE) was performed using voom (Law et.al., 2014) with gene expression counts with associated precision weights, followed by linear modeling and empirical Bayes procedure using limma.

Within each tissue, the following linear regression model was used to detect secually dimorphic alternative splicing event expression:

   y = B0 + B1 sex + epsilon (error)

where y is the gene expression to be modeled sex denotes the reported sex of the subject. The function named fit_tissue() performs this analysis and accepts two arguments, the tissue and an object and create the model matrix based that tissue's sex. We will perform a linear fit after calculating normal factors (based on the library size) and calculate the dispersion using voom (mean variance model of dispersion). We are saving the resulting matrixes as files.

Content in this release are the results from executing the jupyter notebook, differentialGeneExpresionAnalysis.ipynb found in the github repository in the jupyter subdirectory.

This release contains the output of the execution of this notebook.

1.3 Output

For each tissue, as selected and specified in the tissues.tsv file found in the ../assets directory. The following files are produced:

1.3.1 ../data/tissue_DGE.csv

This file contains the topTable results, reporting the ENSG- gene identification,logFC - log fold change, AveExpr, t - the model result for the sex-bias (see section 2, P.Value, adj.P.Val - bon, and B (FDR).

1.3.2 ../data/tissue_refined.csv

This are the values that are differentially expressed with results that are 1.5 fold change greater than the mean and with a p-value of less than 0.05.

1.3.3 ../data/tissue_ensg_map.csv

This is the mapping using gprofiler of the ENSG identifiers to their geneSymbols for ease of filtering prior to creating the linear model of the junctions in the computational step for the differential alternative splicing is completed.

1.3.4 ../pdf/tissue-gene-y-voom-MDSplot-100.pdf

These are the counts in a multi-dimensional scaling plot (MDSplot), showing the ability for the model to segregate the sex as illustrated with red m for the male and blue f for the female self-reported sex phenotypes. In these plots, voom has been used to model the variance.

1.3.5 ../pdf/tissue-gene-y-MDSplot-100.pdf

These are the counts in a multi-dimensional scaling plot (MDSplot), showing the ability for the model to segregate the sex as illustrated with red m for the male and blue f for the female self-reported sex phenotypes. In these plots, this is without the application of the results from modeling the variance with voom.