Brielin Brown
@BrielinB
Followers
372
Following
388
Statuses
161
Asst Prof @UPennDBEI @PennGenetics. Interdisciplinary scientist. Bedroom DJ. Aspiring surf bum. He/him.
Brooklyn, NY
Joined October 2011
@SashaGusevPosts This paper has like 2k citations and people seem to completely forget it all the time
0
0
3
0
0
3
@ekernf01 @david_a_knowles @tuuliel_lab @johnomix It is fairly reasonable although off-target effects are definitely a thing. And no, not within this model. That's a much stronger intervention and given the variable effectiveness of gRNA at inhibition we thought the mean shift more appropriate.
0
0
1
@ekernf01 @david_a_knowles @tuuliel_lab @johnomix So rather than assume the intervention changes the graph structure, it just changes the mean value of it's target by some amount. A drawback is we dont model uncertainty in the target (X is assumed known).
1
0
1
1
0
0
@AshkaanF @lpachter So unlike genetics where large changes in af x ancestry dominate, in other oms theres tons of small individually insignificant differences that get buried in noise. CCA allows you to ask - what's in common across these data? And again the dominant signal turns out to be ancestry
0
1
3
Yes, comparison to other methods + more analyses of the hidden spaces looking carefully at the factor loadings. Altogether this is a great example of how data integration can be used to generate plausible mechanistic hypotheses linking genetics -> disease!
Congrats @BrielinB - great to see this work out! Compared to the preprint we added comparisons to other methods, showing that MFCA works really well in capturing biological variation. Give it a try!
0
0
0
A big thanks to all the TOPMED/MESA investigators for helping us get this out, and to @lpachter and @yarbsalocin for inspiring me to think deeply about CCA many years ago ( (4/4).
0
0
7
RT @david_a_knowles: First time at #ASHG with the lab! Tonight @BrielinB has a poster spotlight in the second session (7-8.15pm): "Multiset…
0
3
0