look at the "tissue-specific" eQTL highlighted in gtex pilot paper

[stephens999]

we should look at our results for that eQTL, see what they look like. The results are in Figure S14, and the SNP is rs37055 the gene is NDRG4 and the pilot data suggest this snp is much stronger eQTL in heart than other tissues. If I remember correctly, because of strong sharing among tissues, it still had an appreciable posterior prob of being an eQTL in other tissues. We should look at the lfsr in Sarah's analysis.

[surbut]

Do you have the summary stats for this particular pair readily available? If so I can compute the relevant posterior quantities promptly.

[surbut]

The GTeX portal only has summary statistics for this in heart left ventricle, and I think we may need SQL (or some search algorithm) to search the list of batches of summary statistics to find the summary statistics for this gene-snp pair across tissues.

[stephens999]

I would first grep this gene in the files that list the genes in each batch. That will tell you which batch it is in. Then grep that snp in the summary stats for that batch

[stephens999]

this gene is http://useast.ensembl.org/Homo_sapiens/Transcript/Summary?t=ENST00000356752;r=16:58500143-58511735;g=ENSG00000103034;redirect=no

ENSG00000103034

[mstephens@midway-login1 lists_genes]$ zgrep ENSG00000103034 *.gz list_genes_077.bed.gz:16 58496749 58496750 ENSG00000103034.10 1000 +

so it is in genes_077

[stephens999]

hmm but: zgrep rs37055 *077* produces nothing in the sumstats directory.

[stephens999]

I can find that SNP in that batch though: zgrep rs37055 * list_snps_077.txt.gz:rs37055

[stephens999]

@surbut so I think maybe eQTLBMA is skipping that snp in summary statistics computation? Is there a log file indicating which snps were filtered?

[stephens999]

according to http://useast.ensembl.org/Homo_sapiens/Export/Output/Location?_format=HTML;db=core;focus=variation;output=ld;pop1=1000GENOMES:phase_3:CEU;r=16:58509444-58529443;v=rs37036;vdb=variation;vf=111381904

it is in complete ld with rs37034 in CEU

[stephens999]

that one is in our summary statistics. I extracted the data using zgrep rs37034 *077* | grep ENSG00000103034 > rs37034.txt but by eye not sure it looks like it is associated... should maybe also check this one on gtex portal

[stephens999]

It is associated on portal. http://www.gtexportal.org/home/eqtls/calc?tissueName=Heart_Left_Ventricle&geneId=ENSG00000103034.10&snpId=rs37034

[surbut]

I don't have access to the cluster currently but I kept track of excluded snps on the uvlr_logs_april folder ? (Don't remember full path) in the analysis directory

Sent from my iPhone

On Jun 5, 2015, at 9:43 AM, stephens999 notifications@github.com wrote:

@surbut so I think maybe eQTLBMA is skipping that snp in summary statistics computation? Is there a log file indicating which snps were filtered?

— Reply to this email directly or view it on GitHub.

[kkdey]

the tissue specific eQTLs for blood, testis, lymphocytes and brain, and tissue consistent eQTLs, shared across all tissues are presented in the file: https://github.com/stephens999/gtex-viz/blob/master/gtex.Kushal/src/tissue_specific_effects_plot.Rmd

[kkdey]

Now working on linking these eQTLs to their gene expression

[surbut]

I added https://github.com/stephens999/gtex-viz/blob/master/Plots/ENSG103034.Rmd a plot of the gene highlighted in the pilot paper. The Matrix EQTL raw statistics don't look particularly strong in any of the other tissues (they are 5.867008 in Heart Atrial Appendage and 1.028463 in Heart Left Ventricle) and the lfsr is 0.1396725 and 0.1450976 respectively. It has high lFSR in all the other tissues. I added my plots as well as my shrunken posterior mean, lfsr and the original table Gao Produced in https://github.com/stephens999/gtex-viz/tree/master/data/matQTLResults and my plots are as listed above. I added a few other snps for comparison because QTL charts doesn't like plotting one snp at a time.