Machine learning, RNA transcription, genomics, statistical genetics. Postdoc @StanfordMed with @anshulkundaje. Previously PhD student with @charlesdanko
Palo Alto, CA
Joined July 2013
- Tweets364
- Following607
- Followers239
- Likes348
Pinned Tweet
Super happy to share that the first paper from my postdoc is now out on bioRxiv (biorxiv.org/content/10.64898…). We trained ProCapNet models on all ENCODE PRO-cap experiments (>200 datasets), representing all major human organs/tissues.
A short application note describing pydreg and benchmarking the GPU implementation is out now on bioRxiv!
biorxiv.org/content/10.64898…
Happy to announce that I (+Claude/Codex) have successfully ported/rewritten the popular dREG peak calling method for nascent transcriptomics (GRO/PRO/ChRO/mNET-seq). code: github.com/adamyhe/pydreg, tweetorial below:
Happy (and utterly exasperated) to announce that I (+AI) have successfully rewritten and optimized LDetect (~60x faster), an old but still commonly used method for dividing a genome into approximately independent linkage disequilibrium blocks. Code: github.com/adamyhe/LDetect-l…
Happy to announce that I (+Claude/Codex) have successfully ported/rewritten the popular dREG peak calling method for nascent transcriptomics (GRO/PRO/ChRO/mNET-seq). code: github.com/adamyhe/pydreg, tweetorial below:
We vibecoded a Slack app that detects bioRxiv and medRxiv links posted to our lab's #papers channel, and auto-posts the title+authors+abstract for easy previewing
Posting the Github repo here - Free to install/deploy in case anyone else finds it useful
github.com/rsatija/biorxiv_p…
1
140
biorxiv.org/content/10.1101/…
Hey we're finally PrediXcan! ... on genes in the training set 😅
1
54
Does training sequence-to-function models on matched functional & genome sequences improve variant effect prediction? For local functions like transcription initiation/enhancer activity, our view is yes. Please check out our new preprint: doi.org/10.1101/2024.10.15.6…
This was joint work with @NathanPalamutt3 (a Cornell undergrad!) and the ever patient @charlesdanko. Will post a full blurb when I have some more free time, but if you want a short elevator pitch, my spotlight presentation at MLCB this year is on YT: youtu.be/DgLlPo6dVV0?si=3EyX…
1
2
733
That feeling when you realize you've been using exponential linear instead of exponential activation this whole time. 😭😭😭
1
96
Dear ML/CS people. What exactly goes in a 2 page/extended abstract? Do I need a mini-abstract/summary at the top? How many figures are expected?
-a confuzzled biologist.
99
biorxiv.org/content/10.1101/… OK, I don't know if this is an ESL issue, but seriously?? "Machine learning warrior"?? "Algorithm legion"?!!
1
2
108
Excited to finally share my work with @charlesdanko on the regulatory syntax governing transcription initiation and training genomic DL models on diverse genomes. doi.org/10.1101/2024.03.13.5…. Blurb thread below: 0/8
First, want to point out that this topic has gotten quite popular! @KDudnyk47866 and @zhou_jian posted Puffin last year, and now there’s CLIPNET (us) and ProCapNet from @kellycochra and @anshulkundaje (to come shortly). Please check out their work too! 1/8
1
2
308
@qz Should be 3000 years ago during the Shang, not the N. Song (~1000 years ago) qz.com/1087121