Rajeeva Lochan Musunuri

@AravSrinivas The subtle difference between a "Research Max" vs "Labs Max" query is a bit confusing (#0" target="_blank" rel="nofollow noopener">perplexity.ai/search/whats-t…). My understanding and experience so far has been that "Labs Max" is much better. Is there any scenario where "Research Max" is preferable over "Labs Max"?

Any Perplexity query on Comet is a lot faster than doing it on another browser. We have optimized it pretty well. And will continue to do so.

Very happy to officially share the high-coverage 1000 Genomes Project (1kGP) whole-genome sequencing (WGS) resource, out today in @CellCellPress! cell.com/cell/fulltext/… (1/7)

My team @nygenome is growing. Looking for a highly motivated, detail-oriented Bioinformatics Programmer to help us develop novel tools for whole-genome sequencing data analysis. Please spread the word: nygenome-openhire.silkroad.com/epostings/inde…

From @lh3lh3 et al., if you were waiting for a description of what minigraph (lh3.github.io/minigraph) is doing that's higher-level than the C code :) . Looks like an exciting read!

Long walk to genomics: History and current approaches to genome sequencing and assembly sciencedirect.com/science/articl…

Long read sequencing of 1,817 Icelanders provides insight into the role of structural variants in human disease disq.us/t/3js36pw

CNVs should be classified consistently between patients based on evidence, with a discussion of clinical significance for individual patient in the report say ACMG & @ClinGenResource in new technical standards go.nature.com/2CjrvhO #CNVs #medicalgenetics @AMPath @GeneticCouns

@lizworthey: Clinical review of 200+ variants per patient takes about 3 hours. Variant fatigue is real! This needs to be faster #gi2019

@brent_p Probably has something to do with this? lemire.me/blog/2018/09/0…

somalier does a lot of bitwise operations and popcounts to calculate IBS0 and other kinship values. I did an experiment to see if porting to avx512 would speed it up: gist.github.com/brentp/11a497d… it does not. probably need to do the popcounting outside of the loop. any other ideas?

1/ shapley values (and the Python shap package) have become an *integral* part of my machine learning methodology I’m surprised by how many people still don’t know about them github.com/slundberg/shap

the oom (out of memory) killer

Bioinformatician: Annotate all the data. Insights will be obvious! Statistical geneticist: Cluster and plot big data. Insights will be obvious! Software Engineer: Make an API to the data. Insights will be obvious! PhD Advisor: Stare at data. Insights are obvious!

Jason Chin @infoecho introduces hierarchical minimizers and minimizer pair binning to compute human genome assemblies in < 2 hours from high accuracy long reads. simple and beautiful. best #SFAF2019 talk for me!

Our new work on understanding the architecture of complex traits "Trans effects on gene expression can drive omnigenic inheritance" is out today cell.com/cell/fulltext/… Great collaboration with @XuanyaoLiu @yang_i_li

Population structure in the age of large-scale genotyped studies em.rdcu.be/wf/click?upn=l…

Let's talk about references and citations. References help readers discover relevant work and reproduce results. 1/N

There's a ton of pedantry in programming. This may be confusing until you realize that programmers spend 8+ hours a day working with the world's most irritating pedant -- the computer. And the computer is .. kind of an a-hole to be honest.

Our review on #DeepLearning for Genomics is finally out - lot's of paper-digging led by @gokcen and @avsecz from @gagneurlab. Targeted at biologists who want to learn how 'AI' can be used for genomic modeling, from motif discovery to single cell genomics. nature.com/articles/s4157…