tnieuwe / gtex_variation_atlas

Determining not only what variation exists in GTEx, but what drives said variation.

GTEx_Variation_Atlas

Overall goal of project

Determining not only what variation exists in GTEx, but what drives said variation.

Data acquistion:

• gtex_read_counts.gct* The version 8 readcounts, are downloaded through R scripts, for most recent download see gtex_read_counts.gct last date modified in Setup Code/downloaded_input/gtex_read_counts.gct
• gtex_phenotypes_v8: A file containing deeper phenotype data on GTEx participants was originally acquired through dbGAP on 2019-10-21
• GTEx_Analysis_v8_Annotations_SampleAttributesDS.txt: a file containing sample data on GTEx samples was last download 2019-9-30
• GTEx_Analysis_v8_Annotations_SubjectAttributesDS.txt: A file containing subject data on GTEx subjects was last downloaded 2019-8-26.

File descriptions

gen_script_gtex_v8_1:

1. Separates tissues readcount matrices from one another.

gen_script_gtex_v8_2:

2. Normalizes the data using DESEq2's VST() function before filtering genes on a threshold that rowMeans() should be >5 normalized counts.

gen_script_gtex_v8_3_v2:

mccall_analysis_corr_step.R:

3. The gene data is further filtered to only genes in the .98 percentile of high variance genes for each tissue.
4. Then using a Kendall's tau correlation the high variance genes generate a correlation matrix.

mccall_cluster_generator.R:

5. This correlation matrix is used to create euclidean distances to cluster the data on absolute values. Generate actual clusters through a kendall's tau critical value.
6. Further break up clusters based on how tightly they correlate using a quantile threshold of correlation, and new agglometric clustering method.

sub_cluster_maker.R:

7. Break the clusters up on if they positively or negatively correlate with one another.

mccall_profile_maker.R:

8. Using the genes from each cluster, generate a Z-score of samples association with each cluster using the normalized count data.