search

hbctraining / DGE_workshop_salmon_online

https://hbctraining.github.io/DGE_workshop_salmon_online/
163 stars 75 forks source link

GSEgo error #18

Closed mistrm82 closed 2 years ago

mistrm82 commented 3 years ago

Error about no term enriched.

Should we look into this?

rkhetani commented 2 years ago

Note about using a higher numbers of permutations.

mistrm82 commented 2 years ago

Tried running the code as is in the lesson but I get the no terms enriched. Setting nPerm=2000 or even 5000 is not noticeable in terms of running time, and it results in a set of terms that are now enriched.

How did terms go from being not significant to significant? The nPerm parameter specifies how many times this randomization is done and more randomizations are performed, the more precise the FDR Q value estimation will be. So at 1000 a good chunk of terms all get the same estimate of 0.07. With more permutations you will find that the individual terms will obtain more accurate values rather than binned shared value.

Problem is, now the term that we use in the lesson for the GSEplot is not showing up. My results have changed.

Solution, remove the GSEgo section and just add a note. Go terms are better assessed with ORA

mistrm82 commented 2 years ago

updated lesson