Daniel Baker

@dnb_hopkins

I'm a computer science PhD student at Hopkins, where I develop fast, memory-efficient methods. he/him/his

Vrijeme pridruživanja: rujan 2017.

Tweetovi

Blokirali ste korisnika/cu @dnb_hopkins

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @dnb_hopkins

  1. proslijedio/la je Tweet
    27. sij

    Conterminator terminates contamination in genomes. and me report over 114K/2M contaminations in RefSeq/GenBank and two unexpected ones in GRCH38 alt. scaffold and C. elegans ref. genome. Preprint: Code:

    Poništi
  2. proslijedio/la je Tweet
    20. sij

    has always been infuriatingly uninformative and unpredictable. We have now, predictably, made it incredibly informative. We present: the RADIOHEAD PUBLIC LIBRARY.

    Poništi
  3. proslijedio/la je Tweet
    14. sij

    My blog post from 2017 which I converted to a bioRxiv pre-print last year has now been peer reviewed and published in Nature Biotechnology : "Droplet scRNA-seq is not zero-inflated"

    Poništi
  4. 9. sij

    Today I found that the blaze lib's CustomMatrix/CustomVector classes work transparently on mmap'd data, which lets you work on impossible huge matrices without rewriting code.

    Poništi
  5. 6. sij

    Call me Rakim, 'cuz I got max flow

    Poništi
  6. proslijedio/la je Tweet
    20. pro 2019.

    Today, published an exposé by Rodrigo Ochigame (a PhD candidate at MIT) on the invention of "ethical AI" by big tech. Here's why I think it's important to read this piece and share it widely. 1/

    Prikaži ovu nit
    Poništi
  7. 20. pro 2019.

    After seeing Lemire's recent work on xorfilters (), this set of AMQ structures seem worth a look. Some of these trade composability (cuckoo/vacuum) and/or streaming updates (xor) for higher accuracy per bit and/or deletion support.

    Poništi
  8. 5. pro 2019.

    Ben's been a fantastic advisor -- I'm happy to discuss both the program/department here a Hopkins, as well as what it's like to be in tour lab. Feel free to contact me by DM here or email at dnb@cs.jhu.edu.

    Poništi
  9. 3. pro 2019.

    In this way, it's similar to one-permutation b-bit minhashing. The key difference is that we use HLL cardinalities as the intermediary for set operations, which lets us calculate containment and symmetric containment, which those sketches don't natively.

    Prikaži ovu nit
    Poništi
  10. 3. pro 2019.

    Mash uses a bottom-k minhash sketch, which only requires one hash function (vs k hash functions) and calculates set similarity by merging sorted lists. HLL is equivalent to storing the log2 minimizers of one-permutation (aka k-partition) minhash.

    Prikaži ovu nit
    Poništi
  11. 3. pro 2019.

    In anticipation of Dashing's upcoming publication, Ben's put together a commentary on the relation between MinHash and HyperLogLog sketching for fast set comparisons. Feel free to ask questions & discuss!

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    29. stu 2019.

    I know that you think that JavaScript supports 64-bit integers, but it does not. And because it is a safe language, it will never let you concern yourself with such low-level issues, it will round numbers off for you, silently.

    Poništi
  13. proslijedio/la je Tweet
    7. stu 2019.
    Poništi
  14. proslijedio/la je Tweet
    4. svi 2019.

    Great work by , co-advised by and myself and in collab w/ Peter White et al at Nationwide Children's. Single-sample mosaic variant calling with linked reads. (I'm pictured b/c Charlotte couldn't be there for the award). Preprint:

    Poništi
  15. proslijedio/la je Tweet
    12. tra 2019.

    String Synchronizing Sets: Sublinear-Time BWT Construction and Optimal LCE Data Structure, D Kempa, T Kociumaka

    Poništi
  16. proslijedio/la je Tweet
    4. tra 2019.

    If your paper is arXived but you don't tweet it… Proved a No Free Lunch theorem 🚫🍱 for clustering tasks, using our recently advocated metric. No best algo! With and , who survived my quibbles about \mathcal vs \mathbb \vs \mathfrak!

    Poništi
  17. proslijedio/la je Tweet
    26. ožu 2019.

    Inspired by others, we created an "awesome-CRISPR" list to collect tools/webservers/other stuff for genome engineering. Please help us complete this collection! Thanks to Lin and Sitong in my group for their work!

    Poništi
  18. proslijedio/la je Tweet
    19. ožu 2019.

    Our newest method is out: Atlas-CNV. We demonstrated FDR reduction in eMERGESeq samples without sensitivity down to single exons. Propose guidelines + criteria for high confidence single-exon CNVs.

    Poništi
  19. proslijedio/la je Tweet
    17. ožu 2019.

    From Shilpa Garg, , , George Chuch: haplotype aware assembly in pedigrees, apparently more accurate and contiguous than TrioCanu

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    15. ožu 2019.

    xxHash new release v0.7 : the family is expanding with a new member, XXH3, offering dramatically improved speed for both large and small inputs :

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·