Super exited to share the second chapter of my dissertation with the world! TLDR; when we revisit Lewontin’s Paradox of Variation in plants with k-mers cool things happen (1/19) 🧵👀

Nucleotide diversity 🧬 and k-mer-based metrics of genetic diversity across 112 plant 🌱 species, amounting to over 205 terabases of DNA sequencing data from 27,488 individuals https://t.co/Gw5YTPM7Ai #biodiversity #genomics

Our paper on using inferred tree sequences for accurate pop gen inference (by training neural nets on tree seqs *inferred* from simulated data) is now published in MBE: https://t.co/eIktFfNvGh Orig rambling thread on preprint below:

I'm recruiting students / postdocs to join my new lab at the University of Rochester for Fall 2025 onwards! If you're interested in phylogenetic comparative methods, genome evolution, and/or computational biology, please get in touch! More info: https://t.co/YqAB88jbEA

Breaking my social media hiatus to promote mine and @moriahyoung's new paper!! We look at how warming & herbivory interact to affect multiple plant traits, and we point to the need for more long-term studies that include abiotic and biotic stressors 🌱

My lab has 2 postdoc positions funded by #NIH and #NSF to study novel ORFs and translational regulation in Arabidopsis and crops. Please retweet.

Check out our new preprint! We show genetic loci assoc. w/adaptation to rust pops, and rethink reciprocal transplants for studying parasite local adaptation. Local adaptation of both plant and pathogen: an arms-race compromise in switchgass rust

🌱 Happy to share one more paper from my time @SchnableLab with the dream collaborative team @szintri & @Rebecca_Roston! Rhythmicity in lipid responses to chilling stress in panicoid grasses & Arabidopsis is now online @JXBot https://t.co/x8zsCe6hFN #Plant #StressResilience

So cool!!! Towards an unbiased characterization of genetic polymorphism

People. There is software to calculate π correctly (including invariant sites). https://t.co/ftHSJPZ54b https://t.co/L82x05r5sy If you insist on using vcftools, please consider this equally accurate but substantial speedup I found online:

If this work interests you, I am looking for post-doctoral opportunities. I'm looking to graduate Spring 2025 and I'm happy to collaborate on writing a postdoctoral fellowship application. You can read the first chapter of my dissertation here: (19/19) https://t.co/tzfdMmV3NA

I feel like this thread is getting long, so I'm gonna stop here, but there are lots of details in the preprint. I welcome any feedback people have on this work. I am super excited to have this project released out into the world!😁(18/19)

So how much of Lewontin's paradox goes away when using k-mers instead of SNPs? I think it's hard to say exactly how much, but this results suggests that SNPs aren't capturing the full picture. It would be cool to see if this result applies outside of plants too (17/19)

This plot is a phylogenetic partial regression, so the relationship that remains after controlling for evolutionary history, genome size, mating system variation, life cycle habit (annual, perennial, etc.), and cultivation status. (16/19)

Here's a main result: nucleotide/SNP diversity hardly changes with population size proxies (i.e. Lewontin's paradox). However, if we use k-mer-based diversity, suddenly the relationship is much stronger! (x axis is a population size proxy based on range and body size) (15/19)

I reprocessed all of this data with the same pipeline to limit the influence of differences in bioinformatic processing on the results. (14/19)

So what I did was calculate k-mer-based and SNP-based estimates of genetic diversity across 112 plant species using publicly available data. This amounted to >200 trillion bp of sequencing data across >27,000 individual plants (13/19)

This approach, which I see being mostly used in metagenomic literature, does not include an alignment step. Thus, it captures extra sequences that would typically be filtered out by alignment. (12/19)

Using the same sets of reads one would typically use for SNP-calling, you can instead count the frequencies of k-mers present in sets of reads and use these counts to measure genetic diversity (11/19)

This is where k-mers come in. K-mers are simply sub-sequences of larger sequences (DNA in this case) of length k, where k is an integer. For example, a 30-mer is a 30 bp long sub-sequence within a larger sequence (10/19)

Thus, such regions are not accounted for in referenced-based genetic diversity estimates, biasing them downward. So, does this systematic underestimation of diversity across species have an effect on Lewontin's paradox? (9/19)