Hello,
I wanted to specify the calling of any QTLs in the PsychEncode data. Were the QTLs called 1) jointly from affected and control individuals, 2) from merged data from GTEx? Or are there specific QTL datasets for controls vs affected individuals also available and individually from different study samples?
Lastly, do I understand correctly that the called QTL dataset are readily and freely to be downloaded here (using different filtering methods):
http://resource.psychencode.org/#Derived
Thank you,
Maris
Created by marisalver @marisalver - see this answer from the analysis team.
**Were the QTLs called jointly from affected and control individuals**
Yes, they were called jointly. And by "jointly", we mean that the associated response data (i.e., expression in the context of eQTLs, cell fractions in the context of fQTLs, etc) were pooled in affected and control individuals. However, we should note that disease status (including "healthy" as one of the variable values for "disease status") were used as known covariates in calculating QTLs, thereby controlling for differences in disease states. So yes -- the QTLs were called jointly from affected and control individual, while taking disease status into account via covariates.
**Were the QTLs called jointly from merged data from GTEx**
This question is somewhat unclear. To be specific, the QTLs were calculated by pooling samples from multiple different datasets. Specifically, we processed data from nine studies: UCLA-ASD, Yale-ASD, BrainGVEX, the The Lieber Institute for Brain Development (LIBD), GTEx, the CommonMind Consortium (CMC), the CMC?s NIMH Human Brain Collection Core (CMC HBCC), and Bipseq of Bipolar cohorts. (See section 9.2 of the Supplement for more details; further details are provided in Section S5 of the supplement to the paper). Thus, GTEx data comprises only a subset of the data used in calling QLTs (and of course, only a subset of the GTEx data was used -- of course, irrelevant tissues in GTEx were excluded).
**Are there specific QTL datasets for controls vs affected individuals also available and individually from different study samples?**
The answer to this is the opposite to that in the answer of the 1st question above: No -- there is not a separate set of QTLs for the different sets of individuals (ie, affected vs. healthy, or for different datasets). If we had done that (which, again, we did not) we would have had substantially reduced power in detecting QTLs. But again, covariates were used for both disease status as well as for study center (ex: Bipbseq vs GTEx vs. LIBD, etc). Thus, these different samples are indeed controlled for.
**Lastly, do I understand correctly that the called QTL dataset are readily and freely to be downloaded here (using different filtering methods):**
http://resource.psychencode.org/#Derived
Yes, absolutely. If you visit that URL, the QTLs are available in the sub-section titled "QTL Maps" under Derived Data Types