I am a computational genomics student who has recently joined a research lab as an RA. We often work with oncoprints (comuts/heatmaps), and we visualize molecular signatures by reorganizing the comut. Although this is the first-line approach, what are some statistical ways to identify molecular signatures? I have heard of NNF, clustering, etc, but I also heard a colleague mention using cox analysis (which I thought was for survival analysis?).

My Background: While introductory statistics has been hammered into my head, I have less experience with higher level computational techniques, like PCA, clustering. I am familiar with the general purpose of the statistical approaches I've mentioned, but not confident in when or how to use them. For example, I know the general purpose of PCA, but I did not know that if you have categorical binary data and want to do unsupervised clustering, you should do logisticPCA. I would appreciate any advice, general or relevant to the question at hand. I am eager to be competent, confident, and contribute to my lab! Thank you :)

  • 1
    $\begingroup$ Can you tell us more about what you are trying to accomplish? Or link to examples? For example I see that comut and oncoprint are specific and distinct visualization softwares (github.com/vanallenlab/comut, bioc.ism.ac.jp/packages/3.2/bioc/vignettes/ComplexHeatmap/inst/…), Cox was a prominent statistician who worked on many techniques, not only survival analysis, but you seem to be using the words in very generic ways so I worry you mean something different. Can you be more specific? $\endgroup$ Aug 15, 2021 at 21:13
  • $\begingroup$ I agree with Max, I have hard time understanding your question. You don't need to explain your presonal context, but rather clearly explain what exactly you do (perhaps with links), what you identify as a problem (why you can not continue doing so) and how do you imagine to solve it. $\endgroup$ Aug 19, 2021 at 17:17


Your Answer

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