jeffrey spence

@spence_jeffrey_

postdoc at stanford working mainly in population genetics.

Vrijeme pridruživanja: travanj 2017.

Tweetovi

Blokirali ste korisnika/cu @spence_jeffrey_

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

  1. proslijedio/la je Tweet
    22. sij

    I'm looking for a computational postdoc for analyzing palaeogenomic data, co-advised with , to start around September 2020. Deadline for applying is March 15th. Link here:

    Poništi
  2. 21. sij

    It's great to see careful consideration of the invariances present in popgen data and rigorous comparison of different architectures. It looks like ABC with summary statistics is still competitive, though, and that there's room for improvement by combining DL and ABC approaches.

    Prikaži ovu nit
    Poništi
  3. 21. sij

    Nice work by 's group using deep learning in population genetics:

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    18. sij

    Our preprint on ancestral haplotype reconstruction is out (and associated code "thread" ). Feedback welcome!

    Poništi
  5. proslijedio/la je Tweet
    16. sij

    Ever wonder if other people dream of flying? Help us find out using our questionnaire

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

    Thanks to the heroic efforts of we have re-released our TAPE code as a pytorch repository! Take it for a spin At NeurIPS? Come chat with us about proteins on Thursday at our poster!

    Poništi
  7. 5. pro 2019.

    Interesting math fact: take a PSD matrix and take the absolute value of every entry. In dimensions 2 and 3 the result is guaranteed to still be PSD, but for any dimension greater than 4 there exist matrices where the resulting matrix is no longer PSD 😱

    Poništi
  8. proslijedio/la je Tweet
    26. stu 2019.

    iSMC is finally here! The much improved version our paper with and is out in . We develop a new model of the recombination landscape and show that it can infer recombination maps with a single pair of genomes:

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    17. stu 2019.

    Felix Wu's work comparing sex-specific germline mutation rates in baboons to humans:

    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    7. stu 2019.

    We all know the written history of Roman civ. With an awesome resource of DNA from 10k+ years and cutting-edge popgen analysis, , , , et al. articulate a new, genetics-based perspective of Roman history/pop migration patterns

    Poništi
  11. proslijedio/la je Tweet
    4. stu 2019.

    Can we learn the language of proteins? Check out our blog post where we discuss Sesame Street, protein plagiarism, and manage to fit in some machine learning along the way.

    Poništi
  12. proslijedio/la je Tweet
    31. lis 2019.

    Our work using geostatistical modeling with ancient DNA to reconstruct ancient migrations in Europe:

    Poništi
  13. 31. lis 2019.

    What's the etymology of the "Tower Property" of conditional expectation in probability? Is "Tower" a person or does it refer to the building?

    Poništi
  14. 23. lis 2019.

    I should also say that it's fast, can handle unphased data, and has some utilities to compute the expected value and percentiles of the distribution of r^2 for SNPs at different genetic distances for a given demographic history.

    Prikaži ovu nit
    Poništi
  15. 23. lis 2019.

    Fine-scale recombination rates are complex traits, likely influenced by many cis and trans factors.

    Prikaži ovu nit
    Poništi
  16. 23. lis 2019.

    In particular, chromatin states characterized by H3K27me3 have particularly elevated recombination rates, and the mechanism for that is not clear. In any case, it's clear there is much more to the story than just PRDM9 binding.

    Prikaži ovu nit
    Poništi
  17. 23. lis 2019.

    The last thing we looked at was whether there are other determinants of recombination rate beyond PRDM9, and we found that chromatin structure appears to modulate recombination rates by about an order of magnitude, but the effect doesn't line up with "open" vs. "closed" chromatin

    Prikaži ovu nit
    Poništi
  18. 23. lis 2019.

    We found some evidence of this, but the effect was much weaker than we had originally expected. Some of this might be because we used binding motifs instead of known PRDM9 binding sites, but there are also a lot of hotspots and not very many recombinations per generation.

    Prikaži ovu nit
    Poništi
  19. 23. lis 2019.

    There's also an interesting story that the double strand breaks involved in recombination events are repaired via homology directed repair, which should cause mutations that disrupt a PRDM9 binding site to increase in frequency.

    Prikaži ovu nit
    Poništi
  20. 23. lis 2019.

    PRDM9 is a protein that positions recombination hotspots, and the C allele of PRDM9 is present at appreciable frequencies only in African populations. Correspondingly we find increased recombination rates at PRDM9-C binding sites only in those populations.

    Prikaži ovu nit
    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:

    ·