David Laub

16 posts

David Laub

David Laub

@dvdlaub

🎓 Bioinfo PhD @ UCSD | 🤖 Deep Learning | 🧬 Cancer Genomics (he/him)

San Diego, CA Katılım Mayıs 2021
287 Takip Edilen57 Takipçiler
David Laub retweetledi
James Talwar
James Talwar@jvt_master·
“Hello there!” 👋 A tad late, but worth the wait! GRIEVOUS, your command-line general, is now published in Bioinformatics! Curious how GRIEVOUS can boost your research? Unravel this thread to become a SNP lord. #Bioinformatics #Genomics #GWAS #PRS 1/9 doi.org/10.1093/bioinf…
English
1
2
2
436
David Laub
David Laub@dvdlaub·
@jmschreiber91 I also wondered whether the mapping array might improve performance, so I implemented a best-of-both-worlds gufunc that used one instead of an alphabet array. I think the results are clear: the mapping array is beneficial so I'll be updating seqpro's implementation soon!
English
0
0
0
25
David Laub
David Laub@dvdlaub·
@jmschreiber91 Using a gufunc (1) improves the flexibility, enabling broadcasting to work with arbitrary n-dim arrays rather than just 1d arrays. I also benchmarked a multithreaded version of tangermeme, but note that afaik multithreaded Numba functions don't support raising errors.
English
1
0
0
26
Jacob Schreiber
Jacob Schreiber@jmschreiber91·
Added a super-fast one-hot encoding function to `tangermeme` last Friday, and I'm still surprised by how fast it is. Timings are encoding chr1. for-loop: ~40s numpy-vectorized: ~12s new: ~1s Thought I'd share some intuition for why it works so well. #L189" target="_blank" rel="nofollow noopener">github.com/jmschrei/tange…
English
1
9
66
12.1K
David Laub retweetledi
Fernan Aguero
Fernan Aguero@fernan_aguero·
Do not use @BioRender. Do not let abusive practices spread! There is an alternative! Use & Contribute to @simonduerr 's open Bioicons! bioicons.com Illustrations are free to use and reuse, and they are available in SVG vector and PNG format without fees!
English
11
290
1.2K
145.1K
David Laub retweetledi
Berk Ustun
Berk Ustun@berkustun·
📢 Please RT 📢 I am recruiting PhD students to join my group at UCSD! We develop methods for responsible machine learning - with a focus on fairness, interpretability, robustness, and safety. Check out berkustun.com/join/ for more information.
English
5
124
334
67K
David Laub retweetledi
Hratch M Baghdassarian
Hratch M Baghdassarian@hmbaghdassarian·
We've put together a protocol and set of tutorials to help those interested in cell-cell communication analysis approach it in a generalizable and user-friendly way. Involved work by some awesome people @eagut and @DanielBDimitrov
Saez-Rodriguez Group@saezlab

1/ 📢 New protocol paper for #CellCellCommunication! Our manuscript presents a pipeline for robust and generalizable cell-cell communication analysis of complex single-cell datasets, combining the best of LIANA & Tensor-cell2cell! 💥🔬 #CompBio #SingleCell doi.org/10.1101/2023.0…

English
1
3
11
1.3K
David Laub retweetledi
Bojing Jia
Bojing Jia@BojingJia·
💃Dancing in the spotlight!🕺Our new paper shows, hidden in plain sight... ...replicated chromosomes, intimately intertwined, in tango!👯‍♀️ Check out the full choreography out today in @NatureBiotech nature.com/articles/s4158…👈🏼
English
11
28
130
37.9K
David Laub retweetledi
Graham McVicker
Graham McVicker@GrahamMcVicker·
UCSD graduate students are integral to research at the Salk Institute. A group of Salk faculty have signed a letter supporting their effort to obtain fair compensation that allows them to live near the university. docs.google.com/document/d/1hV…
English
0
22
69
0
David Laub retweetledi
Andrea Castro
Andrea Castro@andrea_b_castro·
So excited to share our latest research! Here we show that accounting for where a peptide comes from within a cell can help improve elution and immunogenicity prediction
The EMBO Journal@embojournal

Which tumor-derived peptides can serve as #neoantigens for #immunotherapy? Andrea Castro, Hannah Carter et al @UCSDHealthSci @UCSanDiego find that subcellular location matters for MHC elution and can lead to improved prediction embopress.org/doi/10.15252/e…

English
0
3
5
0