That's not adviseable, particularily not when the genes are selected just because of the p-value (and not on some a-priori grounds).

Thanks Jochen. How about if we confirm the important ones with qPCR?

Archita, how do you define "important" if not by your statistical analysis of the RNA sequencing data? If it is by some other means, then why perform the RNA-seq at all?

I agree with Jochen, but would more strongly advise against it. You are asking two questions here, that together you hope would lead you to discovering meaningful biology.

1. Statistical significance? p-value (unadjusted) in this case is much less valuable because it does not account for the multiple testing problem. With respect to this question, report and perform additional analyses (GO, pathways etc) using only those genes that fall within your statistical thresholds (using the q-value). This is where the transcriptomic data shines and you may identify some potential novel biology here. The important thing here is that you do not want a large proportion of your data to be comprised of false positives.

2. Biological significance? Statistics here is only one tool and it alone should not inform your biology. If you have some prior knowledge of the potential significance of a gene or genes that is not borne out in the transcriptomic data, then a rigorous follow up is required (not just qPCR of the same samples you used for the RNA-sequencing). You can certainly report the RNA-seq statistical results of a any gene (even if it falls outside of the thresholds you used for question 1 above), but report both the p-value and q-value, or just the latter. If you feel strongly that this gene is biologically significant, then also report any additional findings supporting the role of this gene in your biological system. If you find supporting evidence, then you will have to discuss the potential reasons why it was not identified as statistically significant in your data. This is where your knowledge of your specific area of research will be critically important.

Best.

Archita, if you know which genes are important, why did you do a screening (NGS, microarrays) then? If the possibly important genes turned out from the screening, just as the "top candidates" (with the lowest p-values or with the highest LFC or because of whatever criterion you chose) and you are able to confirm them with independent samples and a different technology, then the screening results don't matter anymore. The key issue here is that you will frequently fail to confirm your candidates, what costs a lot of time, work, and money (and possibly also the lifes of animals).

Thank you for your responses.

I understand 'important' is the wrong word! What I meant was that there were a lot of genes that had huge fold changes (some even in thousands), significant p-values, but not significant q-values.

If by q-value, you're referring to adjusted p-value based on an FDR correction. If papers can use a 10% (p

HI Archita Ghoshal ,have you figured your problem out ？i faced the same problem with you when I analyzed my RNA-seq data