This is a test version of Biostars. For the public version, visit https://www.biostars.org.
differently expressed genes

Is there any way to check that i have obtained the correct set of differently expressed genes?can i come to know about differently expressed genes by just looking at normalized data? i apologize if some one find my question naive.

rna-seq gene
• 1,567 views • link
updated by Kristoffer Vitting-Seerup Kristoffer Vitting-Seerup • written by maryak • maryak

It is unclear what you have done, which data you have and what you are looking for. Please elaborate.

• 0 views • link
written by WouterDeCoster WouterDeCoster

i have htseq-count TCGA breast cancer data and i have done TMM normalization. basically i want to use machine learning algorithms for my analysis

• 0 views • link
written by maryak • maryak

1 answer

The short answer is no. The reason we are doing DE in the first place is to find those genes :-)

That said:

  • If you are using a well establised, well tested (benchmarked) tool - and your p-value distribution looks good (for more information on this take a look at this blog) there is a good chance you have a good set.
  • Please note that with datasets like TCGA breast cancer (or single cell data) you have so many samples that you have the power to detect very small changes. Therefore I would recomend to also use a cutoff on effect size (in this case the absolute log2FC). This can even be done in the statistical test (testing abs(log2FC) > x instead of the default abs(log2FC) > 0) in a number of tools.
  • Lastly you can as Grant suggest do validations - this can both be experimental such as qPCR or by analysing other similar datasets.

Cheers Kristoffer

• 0 views • link
written by Kristoffer Vitting-Seerup Kristoffer Vitting-Seerup

Log in to answer this question.

More posts like this