mspivakov
commented
1 year ago Hi Martina Thanks for your question and sorry for only spotting it now - we have just realised that the email notifications were going to a defunct email. Is your question still relevant? Please confirm and if so, we'll look into this. Best wishes, Mikhail
Hello!
I am running a differential contact analysis and I am now looking at the Chicdiff outputs. I am struggling to understand what each column represents for both the _results.rds and the getCandidateInteractions() outputs. If I may offer a suggestion, I would recommend adding a column legend in the vignette.
Specifically, for the getCandidateInteractions() output I have the following questions: 1- what is the value shown under each conditionN_replicateX column? (in the vignette it would be columns: NCD4_22 NCD4_23 Mon_2 Mon_3) 2- Given that multiple fragments are pooled together, but the actual chromatin interaction called by Chicago is only one, I don’t understand how I should interpret the output of this function and how to use this for downstream analyses. For each tested oe fragment (oeID column), the function tells me which surrounding pooled fragment is to be prioritised and driving the differential interaction (please, correct me if I am wrong). However, the original tested fragment was the one indicated by the oeID column.. if, for example, I wanted to determine whether the chromatin interaction is lost or gained in different conditions, would it be correct to assign the attributes (i.e. log2fold change, weighted adjusted p-value) of the "driver" fragment to the orginally tested fragment (oeID)? even though the "driver" fragment is not the oeID fragment? And if multiple driver fragments are identified, should I just select the one with lowest weighted adjusted p-value? I hope this is clear.
Thank you very much for any help in clarifying this!