Question

Accounting for Gene Set Overlaps in Goana and Kegga Results

0

Entering edit mode

Dario Strbenac ★ 1.5k

@dario-strbenac-5916

Last seen 4 minutes ago

Australia

Is there a technique to resolve the correlation introduced into gene set results from having gene sets which have similar components? For example, if the top results are

##                                          Pathway   N  Up Down                P.Up P.Down
## path:hsa05169       Epstein-Barr virus infection 178  86   10 0.00000000000000084  1.00
## path:hsa05165     Human papillomavirus infection 289 114   34     0.0000000000012  1.00

it's probably because hsa05169 and hsa05165 share many components.

limma pathway • 1.2k views

ADD COMMENT • link updated 5.3 years ago by mikhael.manurung ▴ 280 • written 5.3 years ago by Dario Strbenac ★ 1.5k

score 1 · Answer 1 · 2019-08-20

1

Entering edit mode

mikhael.manurung ▴ 280

@mikhaelmanurung-17423

Last seen 2.5 years ago

Netherlands

If you are using Gene Ontology, then you could summarise your findings with tools such as REVIGO . The tool will try to find representative pathways based on semantic similarities (i.e. how similar the names are). Or you could also use emapplot in clusterProfiler to visualise the overlapping genesets.

ADD COMMENT • link 5.3 years ago mikhael.manurung ▴ 280

0

Entering edit mode

Thanks for the suggestion. REVIGO works only for gene ontology, so won't be suited to KEGG pathways. emapplot is a good suggestion for visualising the problem but doesn't adjust the statistical modelling to correct for it.

ADD REPLY • link 5.3 years ago Dario Strbenac ★ 1.5k

1

Entering edit mode

Oh, now I understand. Try PADOG instead! I think the name is self-explanatory and might be the one that you're looking for.

ADD REPLY • link 5.3 years ago mikhael.manurung ▴ 280