Entering edit mode
6.7 years ago
fawazfebin
▴
100
Hi I was analysing RNA seq datasets from GEO with latest available pipelines and the latest annotated reference genome. When I compared my edgeR results with the results of the published article , GO analysis results arent similar. The published article had used an older pipeline and an older reference genome for alignment. How do I ensure that my analysis steps are right so that I can go ahead publishing my results of interest ? Thanks in advance.
If you want to exactly reproduce a published analysis, you need to use the same version of everything as in the paper. On the other hand, in my view, what matters most is that whatever biological conclusion reached by the analysis is robust. For example, concluding that treatment A acts on pathway X should not depend on the software version or the genome annotation version otherwise I would question whether this is a real biological effect.
Is it possible to determine the real biological effect with bioinformatic analysis alone ? Thanks .
One has to define what a real biological effect is. In my comment above, I meant it in opposition to a conclusion reached by chance. Most of the time, we are confident an effect is real when it is experimentally reproducible. Re-running an analysis on the same experimental data and reaching the same conclusion tells you that, with this data, the conclusion is reproducible but doesn't give you more confidence this is is a real effect but conversely, as alluded to above, re-running the analysis and coming to a different conclusion is surely an indication that the conclusions are not reliable and so whatever effect is uncovered is probably due to chance.
Great thanks for the views.