Skip to content

11 — Using GSEA

Jump to bottom
narasi15 edited this page Mar 10, 2020 · 9 revisions

Time estimated: 4 hrs ; taken 5 hrs; date started: 2020-03-10; date completed: 2020-03-10

Process

  1. First I installed GSEA following the guide . Before downloading I had to register to access the GSEA software. I downloaded the GSEA v4.0.3 version for Mac App.
  2. Used the code on week 7 lectures slide 35 in R, to download the Baderlab files. I needed to install and load the RCurl package in order for this to work.
  3. Now using GSEA, I loaded both my files (baderlab file & rank list). I clicked on the Run GSEAPreranked option under tools and filled out the correct parameters as suggested (referred to slide 30, week 7 lectures)
  4. The run was successful, and I was redirected to a page with the results by clicking on the result, under GSEA Reports.
Explain the reasons for using each of the above parameters.

Number of permutations: This field provides generally provides a more precise assessment of significance and produces fewer false positives in the data, so setting this really depends of the size of the dataset. So I kept it as the default 1000, since our dataset size was quite large.

Collapse: GSEA allows you to upload a CHIP file to map gene IDs to HUGO symbols, and then use the CHIP file to collapse the dataset. We are not using a CHIP file in this case, so this parameter was set to No Collapse.

Min/Max Size: This parameter tells us which genesets to ignore that contain less that the min number and greater than the max number (according to the GSEA User Guide). To keep consistent with the g:profiler analysis, we decide to use the same values.


What is the top gene set returned for the Mesenchymal sub type? What is the top gene set returned for the Immunoreactive subtype?

The phenotype referring to na_pos is the Mesenchymal sub type, and the na_neg is the Immunoreactive subtype.
Top gene set for each is:
Mesenchymal: "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION%MSIGDB_C2%HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION"
Immunoreactive: "HALLMARK_INTERFERON_ALPHA_RESPONSE%MSIGDB_C2%HALLMARK_INTERFERON_ALPHA_RESPONSE"

What is its pvalue, ES, NES and FDR associated with it.
Mesenchymal:

Immunoreactive:


How many genes in its leading edge?
Based on the images above, we are shown the tags percentage and the gene size. We the calculate the number of genes in the leading edge as such:
Mesenchymal: 193(0.57) = 110.01, so about 110 genes
Immunoreactive: 94(0.7) = 65.8, so about 66 genes

What is the top gene associated with this geneset?
If we click on the GD Details link, in the top row of each subtype, we are lead to the following top genes:
Mesenchymal: FBN1

Immunoreactive: PROCR



Outlooks and Conclusion

  • GSEA was fairly easy and intuitive to use, however it took a long time to run the analysis
  • The lecture slides helped out figure how to operate GUI, and which tabs/buttons to click, and where to fill out the parameters easily

External Resources

(n.d.). Retrieved from https://www.gsea-msigdb.org/gsea/doc/GSEAUserGuideFrame.html
FAQ. (n.d.). Retrieved from http://software.broadinstitute.org/cancer/software/gsea/wiki/index.php/FAQ

This copyrighted material is licensed under a Creative Commons Attribution 4.0 International License. Follow the link to learn more.

Clone this wiki locally