I'm currently working with several RNAseq datasets from the ENCODE database. My approach to assessing the quality of the data, i.e, reproducibility is by automating the assessment of expression profile clustering.

Obviously, increases in sample numbers will force me to automate quantification of clustering performance. However, my lack of knowledge in biology/anatomy leaves me unsure how to approach this problem.

An obvious first step is to ensure cell types cluster together. However, I think that it's also important that similar cell types cluster together. The knowledge I lack is, precisely, which cell types I should expect to have similar gene expression profiles.


What ontology sets or databases or biological approach would you recommend for this task? I'm currently considering merging several ontologies (Cellosaurus, CL and more), but perhaps there is a better approach.

Thanks in advance, hopefully this question is not too biological for this site.

  • $\begingroup$ Hi @Jaslibra, can you perhaps be more specific, are you assessing the quality of the ENCODE data or clustering algorithms specifically, as you mention both? $\endgroup$ – Matt Bashton Aug 2 '18 at 15:53
  • $\begingroup$ @MattBashton, I'm not looking to assess clustering algorithms directly, but this would be a natural consequence of what I'm trying to achieve. Specifically, my question asks: which cell types are expected to be similar? And what resources can I use to automate checking how my clustering meets these expectations? $\endgroup$ – jaslibra Aug 2 '18 at 16:12

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.