I compared two manually defined clusters using Seurat package function FindAllMarkers and got the output:

enter image description here

Now, I am confused about three things:

  1. What are pct.1 and pct.2?
  2. How come p-adjusted values equal to 1? What does it mean?
  3. If we take first row, what does avg_logFC value of -1.35264 mean when we have cluster 0 in the cluster column? Is it that in cluster 0 the Cttnbp2 gene downregulated by a factor of 2^1.35264?
  • 4
    $\begingroup$ satijalab.org/seurat/de_vignette.html $\endgroup$ Commented May 9, 2018 at 6:36
  • 1
    $\begingroup$ You have a few questions (like this one) that could have been answered with some simple googling. I suggest you try that first before posting here. $\endgroup$
    – Devon Ryan
    Commented May 9, 2018 at 9:53
  • 1
    $\begingroup$ I could not find it, that's why I posted. Thank you @heathobrien! Should I remove the Q? $\endgroup$ Commented May 9, 2018 at 18:18

1 Answer 1

  1. pct.1– The percentage of cells where the gene is detected in the first group

  2. p_val_adj– Adjusted p-value, based on bonferroni correction using all genes in the dataset.

This is not also known as a false discovery rate (FDR) adjusted p-value.

An adjusted p-value of 1.00 means that after correcting for multiple testing, there is a 100% chance that the result (the logFC here) is due to chance.

  1. You are correct. The expression of Cttnbp2 donwn-regulated by that factor. If you were to look in the table at the cluster 1 DE genes you would see Cttnbp2 with a logFC value of 1.35264. When you are looking at genes differentailly expressed in only two clusters, the sign is the only thing that changes.

Vignette posted above by @heathobrien.

  • 1
    $\begingroup$ How the adjusted p-value is computed depends on on the method used (use.test). E.g. "DESeq2" uses Independent Hypothesis Weighting. While Bonferroni correction does indeed control the FDR as a side effect it is build to control the family-wise error error rate. An adjusted p-value of 1 does not mean that the result is due to chance, it just means that it cannot be excluded (by the methods used) that such a difference would be observed was there no true difference in expression. $\endgroup$
    – jan-glx
    Commented Aug 16, 2020 at 13:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.