Interpreting GSEA results

Question

[Ron]

Hi all,

I am working with GSEA results using GSEA pre ranked. My input is list of genes with their log2foldchange as the metric . The genes with log2foldchange were calculated using Differential expression.

Now in my GSEA results. I see this pathway plot. The pathway name is RB_P130_DN.V1_DN and its description in Msigdb database is "Genes down-regulated in primary keratinocytes from RB1 and RBL2".

So this pathway is upregulated in my experiment or downregulated ? Can someone explain this result ? I see that the genes in my experiment have a positive Enrichment score and the pathway seems to be upregulated to me. But then it does not correlate with the pathway as in that the genes go down ? https://ibb.co/mVWS5R

Thanks, Ron

[abbey]

I've been struggling with interpreting GSEA results the past few days as well. It would be helpful to know what your experiment was, but based on my understanding of GSEA this pathway was upregulated in your experiment.

This gene set is of genes that are downregulated when RB1 and RBL2 are knocked out. If your plot was primarily negative, then that would mean your data looked similar to a RB1 & RBL2 KO. Instead, most of these genes are upregulated in your data, so your experiment is correlated to the presence of those two genes.

[Ron]

Yes , I also think the same . that in their experiment the genes are downregulated when RB1 and RB2 are knocked out, but in our data those genes are up regulated. So can we say that still our data is related to those two genes being knocked out / down ,despite of the fact that the genes regulated by RB1 and RB2 knockdown are upregulated ?

[pshubhamoy]

Hi , I have a formal class (S4) data like this

Formal class 'gseaResult' [package "DOSE"] with 10 slots ..@ result :'data.frame': 20 obs. of 11 variables: .. ..$ ID : chr [1:20] "rno05200" "rno04010" "rno04151" "rno05166" ... .. ..$ Description : chr [1:20] "Pathways in cancer" "MAPK signaling pathway" "PI3K-Akt signaling pathway" "HTLV-I infection" ... .. ..$ setSize : int [1:20] 370 224 231 203 152 133 118 108 102 136 ... .. ..$ enrichmentScore: num [1:20] -0.384 -0.401 -0.388 -0.425 -0.432 ... .. ..$ NES : num [1:20] -1.54 -1.52 -1.48 -1.59 -1.56 ... .. ..$ pvalue : num [1:20] 0.00164 0.00167 0.00169 0.00173 0.00174 ... .. ..$ p.adjust : num [1:20] 0.0145 0.0145 0.0145 0.0145 0.0145 ... .. ..$ qvalues : num [1:20] 0.00836 0.00836 0.00836 0.00836 0.00836 ... .. ..$ rank : num [1:20] 2544 1610 1928 1804 2600 ... .. ..$ leading_edge : chr [1:20] "tags=34%, list=23%, signal=27%" "tags=23%, list=15%, signal=20%" "tags=29%, list=18%, signal=24%" "tags=23%, list=17%, signal=20%" ... .. ..$ core_enrichment: chr [1:20] "171140/294962/25729/314384/29560/367072/288264/399489/116502/24426/24654/170668/501110/81685/29492/287357/11448"| __truncated__ "25266/25054/360640/59323/79114/24446/25267/29496/24674/114495/117269/25597/78965/116683/24516/292763/24329/2571"| __truncated__ "65248/361696/292406/24514/89805/78975/25513/300253/29302/25155/310553/64033/25266/25054/59323/79114/25634/11666"| __truncated__ "308995/84420/303539/414788/313050/365527/114212/64033/25266/360640/84426/299611/25267/84027/24674/24516/25289/5"| __truncated__ ... ..@ organism : chr "rno" ..@ setType : chr "KEGG" ..@ geneSets :List of 324 .. ..$ rno00010: chr [1:72] "100145871" "100364027" "100364062" "100911515" ... .. ..$ rno00020: chr [1:33] "100125384" "103690168" "103693780" "114096" ... .. ..$ rno00030: chr [1:31] "100360180" "108348261" "114508" "24189" ... .. ..$ rno00040: chr [1:34] "113992" "116463" "154516" "171408" ... .. ..$ rno00051: chr [1:39] "100364027" "100911515" "100911725" "114508" ... .. ..$ rno00052: chr [1:32] "100364027" "103690059" "114860" "116463" ... .. ..$ rno00053: chr [1:27] "113992" "154516" "24861" "24862" ... .. ..$ rno00061: chr [1:14] "113976" "114024" "116719" "117243" ... .. ..$ rno00062: chr [1:31] "100911186" "102549542" "113965" "140547" ... .. ..$ rno00071: chr [1:47] "100145871" "100911186" "113965" "113976" ... .. ..$ rno00072: chr [1:11] "100361036" "117099" "24450" "25014" ... .. ..$ rno00100: chr [1:19] "114100" "114700" "117278" "140910" ... .. ..$ rno00120: chr [1:16] "170588" "192242" "246211" "25284" ... .. ..$ rno00130: chr [1:12] "103693015" "24314" "24813" "25249" ... .. ..$ rno00140: chr [1:84] "100362350" "100910877" "108348086" "108348266" ... .. ..$ rno00190: chr [1:143] "100188937" "100361126" "100361960" "100362331" ... .. ..$ rno00220: chr [1:20] "192268" "24398" "24399" "24401" ... .. ..$ rno00230: chr [1:182] "100360582" "100361574" "100362333" "100363253" ... .. ..$ rno00232: chr [1:6] "114768" "116631" "116632" "24297" ... .. ..$ rno00240: chr [1:104] "100360582" "100361574" "100362333" "100363253" ... .. ..$ rno00250: chr [1:35] "100360621" "117544" "192268" "24240" ... .. ..$ rno00260: chr [1:40] "103691744" "114027" "114123" "171133" ... .. ..$ rno00270: chr [1:47] "100360621" "100912604" "103691744" "171347" ... .. ..$ rno00280: chr [1:56] "100360621" "100361036" "100911186" "113965" ... .. ..$ rno00290: chr [1:4] "25044" "29592" "360816" "64203" .. ..$ rno00310: chr [1:61] "100169747" "100359816" "100361710" "100362634" ... .. ..$ rno00330: chr [1:52] "100912604" "108348083" "114027" "24264" ... .. ..$ rno00340: chr [1:24] "24443" "25375" "25750" "266603" ... .. ..$ rno00350: chr [1:40] "100145871" "100360621" "103694877" "171178" ... .. ..$ rno00360: chr [1:22] "100360621" "103694877" "171179" "24311" ... .. ..$ rno00410: chr [1:33] "100911186" "100912604" "116593" "140547" ... .. ..$ rno00430: chr [1:11] "116568" "156275" "24379" "24380" ... .. ..$ rno00440: chr [1:6] "140544" "286936" "310773" "362713" ... .. ..$ rno00450: chr [1:19] "100911305" "103691744" "24962" "291314" ... .. .. [list output truncated] ..@ geneList : Named num [1:10844] 8.15 6.39 5.33 4.82 4.78 ... .. ..- attr(*, "names")= chr [1:10844] "116463" "313352" "500685" "246253" ... ..@ keytype : chr "UNKNOWN" ..@ permScores : num[0 , 0 ] ..@ params :List of 6 .. ..$ pvalueCutoff : num 0.05 .. ..$ nPerm : num 1000 .. ..$ pAdjustMethod: chr "BH" .. ..$ exponent : num 1 .. ..$ minGSSize : num 100 .. ..$ maxGSSize : num 500 ..@ gene2Symbol: chr(0) ..@ readable : logi FALSE

how do I subset this based on my desired genesets or pvalue? Any idea? thanks

[shweta.johari13]

Hey Ron,

I am facing the same issue right now. Could you please let me know how you did it?

Yours sincerely, Shweta Johari

Answer 1

[anp375]

Normally, IPA handles this. Was the data normalized? I'd say plot all the log2foldchanges for all genes, and then highlight the genes that were assigned to that pathway.

[Ron]

The data is normalized by DESEQ package while doing differential expression.Thanks for your comment ,however thats not what I asked. I just want to know if my results from differential expression analysis - how they relate with this specific pathway based on the enrichment plot ?