McInerney Lab

James McInerney's lab twitter account (Nottingham, UK). Microbiology, evolution, networks, horizontal gene transfer, Fusions. Lab boss: @jomcinerney

[McInerney Lab @mcinerneylab]

New Lab Paper:

Prokaryotic Pangenomes Act as Evolving Ecosystems https://t.co/rM9UyOcAFv

[James McInerney @jomcinerney]

The video explaining (though I really should have used a script) the panGPT program is here: https://t.co/XboTVDZViz

[James McInerney @jomcinerney]

I have released version 0.10a of my AI program, panGPT. panGPT is capable of building large language models from scratch and is suited for the analysis of pangenome datasets, protein, DNA and long molecular sequences. You can download the program at https://t.co/pdj4nEbxtd

[James McInerney @jomcinerney]

I don't know why it pulled the image from the PNAS paper and not the TREE paper. The TREE paper is here: https://t.co/dgBewgTYy1

[James McInerney @jomcinerney]

Opinion piece by Jo Wolfe on our recent paper on pangenome evolution. https://t.co/dgBewgTYy1

Our paper is here: https://t.co/WcxX345J6c

[James McInerney @jomcinerney]

In the fullness of time I will provide tutorials and more/better documentation. The code should be considered BETA and might have bugs. Please respond through GitHub or email [email protected]

Thank you.

[James McInerney @jomcinerney]

What I would say is that you should read up on both transformers, positional encoding and machine learning in general, before blindly using these programs 6/n

[James McInerney @jomcinerney]

You could examine different “languages” within pangenomes. I would be fairly sure that the “language” of bacillus is different to the language of Escherichia, for example 5/n

[James McInerney @jomcinerney]

You could train a model on a very large presence-absence pangenome dataset and then try to generate novel genomes using appropriate prompts 4/n

[James McInerney @jomcinerney]

I make no guarantee of anything about these programs but here is a suggestion for what these programs might do. 3/n

[James McInerney @jomcinerney]

PanGPT will train a model on a pangenome dataset, while panPrompt will use the model to perform “next token prediction”. 2/n

[James McInerney @jomcinerney]

I have spent the last few months writing a simple transformer program that can be used easily on pangenome data. 1/n

GitHub - mol-evol/panGPT: A Transformer for Pangenome data https://t.co/8qEfxTULCt

[Prof Mary J. O'Connell @Evol_Molly]

We are looking for a computational postdoc (36months) to work on our
@wellcometrust
funded grant to understand genomic adaptation and response to heat stress in mammals. https://t.co/f8ALupSoSw… deadline: Jan 11th 2024. Do get in touch:)

[James McInerney @jomcinerney]

shite time to announce a new paper but if ur online…we just published this. So if u like evolutionary biology, microbiology or science, this one’s for you, baby

Contingency, repeatability, and predictability in the evolution of a prokaryotic pangenome https://t.co/oXKH69Wxft

[James McInerney @jomcinerney]

Maria Rosa has written a very nice explainer of this paper. Short twitter/X thread 👇

[John Dudley @johnmdudley]

Define the problem, justify the finance, explain the impact, explain interdependencies and fall-back positions, proofread !!

[James McInerney @jomcinerney]

Probably the last video I’ll make on this theme. A few pointers for those frustrated with getting a “No” from funding agencies. I hope it’s useful & if so, please give it a like and a follow.

Top 5 Reasons Your Research Grant Proposals Don't Get Funded
https://t.co/S1ivW3B4yc

[James McInerney @jomcinerney]

My latest video is a Book Review: Fundamentals of Data Visualization bby @ClausWilke https://t.co/zh9jHDufoG

[McInerney Lab @mcinerneylab]

Bursts of novel composite gene families at major nodes in animal evolution https://t.co/cYMLemu36c

[Peter Mulhair @PeterMulhair]

So happy to see this out 🎉 check out our work on analyzing composite gene evolution across animals!!
It was an amazing project to work on, and an amazing group of people to work with.

[Prof Mary J. O'Connell @Evol_Molly]

Latest paper now up on BioRXiv: How often does the same composite gene (i.e. fusion/fission) emerge independently in animals? Spoiler alert it happens LOTS and more on certain nodes!;-) All thanks to my amazing co-authors for unwavering dedication .. https://t.co/IMQY0RTvpS