2/2) lastz requires to pre-softmask the genomes extensively which is itself very computationally costly, and since it is often a handful of problematic jobs, it seems more efficient to have lastz mask problematic repeats itself? @hillermich.bsky.social have you though about improving lastz?
1/2) #bioinformatics people interested in #genome alignment, has anyone tried to do "adaptive seeding throttling" where after a long run time a lastz job would go into an adaptive mode where it detects that a specific repeat is initiating too many seeds, which would trigger lastz to mask it?
I have been told that in reference to the big confs in my field, the smaller ones are always very inclusive. Having observed social dynamics at those big confs I see where they come from.
I have now heard repeatedly from non-US PIs they stopped to go to American confs because the PIs there being socially exclusive and not including much anyone else in social interactions. I have also seen it, I find it disconcerting to this day.
Wow, incredible shot indeed!
@pratikkatte.bsky.social and I just released Lorax 🌲, a tool for interactive exploration of biobank-scale ancestral recombination graphs (ARGs).
If you’ve ever wanted to scroll across the ancestries of thousands of genomes… this is for you.
Recently we're working with SNPs from whole genome assemblies to estimate ARGs. It's a pain to go from alignment files to vcf, keeping track of masked and invariant sites. So we wrote a snakemake/SLURM pipeline. Hope it's useful to others, and don't hesitate to post issues if there are problems!
The Goldberg lab at UCLA is hiring one or more postdocs. Flexible start date.
We develop methods to study population genetics of humans, our primate relatives, and our pathogens.
www.goldberglab.org/join
Have you ever wondered how many archaic populations contributed DNA to modern humans? We know about Neanderthals and Denisovans, but the fossil and genetic evidence suggests a much more complex history!
www.biorxiv.org/content/10.6...
Excited to share our new preprint! 🍇🧬
We analyzed 639 genomes across 48 grape species to understand how hybridization drives adaptation.
• ~14% of average Vitis genome is introgressed
• Most parallel adaptations are shared via gene flow
• Two "hybrid species" are actually hybrid swarms
For those interested, a new version (v3.1.0) of poolfstat is now on CRAN, including new functions and updates, new stats (Fh, Fd, FdM) implemented in sliding.windows.fstat; and a revised vignette.
[cran.r-project.org/web/packages/poolfstat/i...
Thinking more about it, maybe the complex mutations at high enough frequencies to see them, and thus having any chance to fix later are those that change less amino acids? The complex mutations that change many aas are more deleterious?
Interesting! It would be a great positive control to test BUSTED-MH. My first thought is that the overall dN/dS could go down while the w2 category of positively selected codons in the model could still go up. Since there are more nonsyn than syn sites, surprising to see dN/dS go down?
Indeed the higher number of parameters decreases statistical power, but the positives ones are far more likely to be true positives.
To add one thing: if you have the data to run both classic dN/dS tests and MK tests, choose the latter. dN/dS tests are blind to positive selection in the important, constrained proteome, while MK test will work regardless of constraint, if DFE well estimated.
#evolution #adaptation #popgen
Synonymous substitution rate variation.
5/5) The claims that lack of congruence between dN/dS and MK tests questions their validity are very weak: there are well known, well identified and well studied reasons why that would be the case. Maybe the field should pay better attention to latest sel tests developments, all above is discussed.
4/5) This is the true hidden killer: classic dN/dS tests are very sensitive to positive selection in weakly constrained, fast evolving immune genes. At the same time, the inflation of PN due to likely ongoing balancing selection in the same genes kills the power of MK test to detect excess of DN.
3/5) The classic dN/dS tests only have statistical power when the selective constraint surrounding the positively selected sites is weak. Beyond needing dN/dS>1, also a matter of statistical power depending on the total number of substitutions. MK approaches really shine in the constrained genome,
2/5) First, the dN/dS tests used in the past comparisons output many false positive results because they do not account for synonymous substitutions or complex multinucleotide substitutions. Look at the -MH tests in HYPHY to account for this.
1/5) Re-reading this excellent paper on using mutation-selection balance models to quantify positive selection. The discussed lack of congruence between MK approaches and classic dN/dS test is entirely unsurprising, see thread below.
www.pnas.org/doi/10.1073/...
Ah, les rapports d'activité tellement lourds qu'ils retardent les dites activités, avec des formats ubuesques et la plupart du temps redondants avec les années passées. Aussi devoir rendre ces foutus rapports quand il suffit d'aller voir Google Scholar. La plaie...
Always a great feeling when the new labs in the EEB Department @uofa-eeb.bsky.social each attract multiple fantastic new grad students. Very happy to see younger colleagues succeed.
Great news for anyone using ARGs in non-model species:
www.biorxiv.org/content/10.1...
Happy to share our last study on how age, biological sex and genetics affect not only the extent of antibody responses against viruses but also the specific viral epitopes recognised. Thanks to Axel Olin @EtiennePatin @LabExMI @institutpasteur @cdf1530 @CNRS
www.nature.com/articles/s41...
I am seeking a postdoc to join my group at UCLA -- ideally the candidate would have some experience in either population genetics or microbes/microbiome (computational background needed). We have a range of projects and are happy to tailer to your interests. Please dm/email me if interested.
Are you interested in human DNA from ancient sediments? Do you want to do a PhD in Vienna? We have two open positions in our group! You can learn more and apply here (scroll down to the very bottom, ours are the last two listings): careers.univie.ac.at/en/praedoc/s...
Counting publications is such a poor proxy for scientific productivity. Counting discoveries, with a bonus for discoveries that were not anticipated, and important insights is a much better proxy. I am wondering if there are standardized, systematic ways of quantifying this.
That looks great! To avoid errors due to starting with spotty gene annotations (gene loss vs. not annotated), I would suggest re-annotating every genome used with TOGA2 from the Hiller lab:
github.com/hillerlab/TO...
Many annotations out there are not optimal and TOGA2 has great power and accuracy.
