I am currently doing a project involving gene rank correlations. How we rank our genes is based on the absolute values of the log2 fold change. I.e. a gene which is doubled with respect to the control condition is listed as log2FC=1 a gene at 4xbaseline would be log2FC=2 and a gene at half baseline would be log2FcC=-1. Since we are more interested simply in if there is an effect i.e. we are hypothesis free about whether or not gene under expression or over expression should be considered important we took the absolute values of these genes so hypothetically a gene at half the expression level of baseline are ranked the same as genes at double the expression level of baseline.

My concern is that this list is across all gene types, i.e. we are lumping kinases in with ion channels and axonal binding etc etc. Would certain gene ontologies have different log fold changes simply due to a smaller range of raw expression? And therefore should be normalized within their ontology?

  • $\begingroup$ Ouch, that's going to be tricky. Especially considering that the majority of proteins have multiple, and often very separate, molecular function annotations. You won't find it easy to separate them into clean bins based on their GO MF annotations. See, for instance: ncbi.nlm.nih.gov/pubmed/26054620 $\endgroup$ – terdon Nov 9 '19 at 11:41
  • $\begingroup$ Thats odd. As a neuroscienctist I was totally unaware of this 'moonlighting'. Are there gene ontologies that define protiens according to more fluid boundries? Such as a hierarchical model? $\endgroup$ – Angus Campbell Nov 13 '19 at 4:59
  • $\begingroup$ Not sure what you mean. GO is a hierarchical DAG. Proteins with an annotation, also inherit all the parents of that annotation. In any case, my point was that it probably doesn't make sense to try or expect to be able to normalize across molecular functions. I wouldn't expect such classes to e very homogeneous in anything apart from function. $\endgroup$ – terdon Nov 13 '19 at 9:17
  • $\begingroup$ Forgive my ignorance, I'm hoping to enter a bioinformatics program. My background is neuroscience I don't have a strong maths background. $\endgroup$ – Angus Campbell Nov 13 '19 at 22:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.