Travis Gibson

Our paper demonstrating that within-species warfare interactions are ecologically important on human skin is now published in Nature Micro! www.nature.com/articles/s41...

How a Scientist Followed Curiosity (and Love) to Build an Innovative Tool to Track Changes in the Gut Microbiome Despite how common they are, scientists still don’t fully understand why some people develop allergies while others don’t.

A nice blog post about our work www.nature.com/articles/s41... can be found here mgriblog.org/2025/08/04/c...

Silence on the relevant literature and errors in implementation - Nature Biotechnology Nature Biotechnology - Silence on the relevant literature and errors in implementation

Im working on something and I am reminded of one of the wildest correspondences ever published www.nature.com/articles/nbt...

The paper has the incorrect dynamics for gLV

The mbtransfer paper also has data leakage issues. In many of the results they apply DESeq2 Normalization in a pre-processing step (for all samples in a bulk process) and include samples which are later used as forecasting test samples when doing so.

When MDSINE2 was supplied actual read abundances the authors note “[w]hen the data were not asinh transformed, the mbtransfer model performed worse than either MDSINE2 or fido”

The apparent poor performance of MDSINE2 in the main text of the mbtransfer paper is entirely from misuse of the model and supplying it a table of "read" abundance that are not actually read abundances.

The nonlinearly transformed values also mess up how the latent dynamics (gLV modeling CFU/g in our case) are interpreting the taxa abundances and how they are changing over time.

MDSINE2 has a Negative Binomial noise model for reads. Seeing transformed "reads" with values between 1 and 10 tells the model that the noise is through the roof for all taxa!

MetaPhlAn 4.2.2 release (initial long-read sequencing support and database update) Announcement We are pleased to share that MetaPhlAn 4.2.2 is now available, which incorporates taxonomic profiling of long-read metagenomes for the first time and includes a new version of the MetaPhl...

The new version of MetaPhlAn v4.2.2 released!

📌taxonomic profiling of long-read metagenomes for the first time
📌new version of the MetaPhlAn db (vJan25_202503) containing >21k new SGBs

Try it out & @hutlab.bsky.social and we are looking forward to your feedback!
forum.biobakery.org/t/metaphlan-...

Our paper is now open access as well.

Incorporating scale uncertainty in microbiome and gene expression analysis as an extension of normalization - Genome Biology Statistical normalizations are used in differential analyses to address sample-to-sample variation in sequencing depth. Yet normalizations make strong, implicit assumptions about the scale of biologic...

New paper in Genome Biology!

genomebiology.biomedcentral.com/articles/10....

We introduce scale models, a generalization of normalizations that explciitly account for uncertainty in biological system scale (e.g., microbial load).

This month's editorial is an invitation to submit papers on clinical and translational aspect of the microbiome.

This collection is a joint effort with @naturemedicine.bsky.social @natcomms.nature.com and @commsbio.nature.com

#MicroSky #MicrobiomeSky 🦠⚕️🧪⚕️

www.nature.com/articles/s41...

Longitudinal profiling of low-abundance strains in microbiomes with ChronoStrain - Nature Microbiology ChronoStrain accurately profiles low-abundance strains in longitudinal samples by jointly modelling nucleotide sequencing errors, strain presence/absence and the temporal information associated with e...

🚨 Out now!

Chronostrain - a method to profile low abundance strains in longitudinal microbiome samples 🦠💻

@travis.gibsonlab.io @younhk.bsky.social @gibsonlab.io @harvardmed.bsky.social @broadinstitute.org

#MicroSky 🦠 #MicrobiomeSky

www.nature.com/articles/s41...

reviewer 3 or 2?

Longitudinal profiling of low-abundance strains in microbiomes with ChronoStrain Nature Microbiology - ChronoStrain accurately profiles low-abundance strains in longitudinal samples by jointly modelling nucleotide sequencing errors, strain presence/absence and the temporal...

"Longitudinal profiling of low-abundance strains in microbiomes with ChronoStrain" - Kim et al.

rdcu.be/ekTwf

It is a great pleasure to listen to @seppekuehnlab.bsky.social
His talk is about ´Learning microbiome design principles from natural variation ´.

Below, the #liveSketch painting during the seminar and given at the end!

#ArtAndScience #MediationScientifique

Several Postdoctoral Fellow openings. All fellows have triple appointment at HMS,BWH, and Broad

- Biological sequence models (theory, design, and application)
- Learning single cell dynamics
- Bacteriotherapy design, “bugs-as-drugs” (using control theory principles)

gibsonlab.io/openings/

We are still accepting registrations for this event! Please come join us and hear about cutting edge microbiome research.

Planning and describing a microbiome data analysis - Nature Microbiology We provide guidance on the planning, execution and description of statistical analyses in microbiome studies.

A must-read for any microbiome researcher 🦠📐and a new addition to our #bestpractices series

Planning and describing a microbiome data analysis by @amydwillis.bsky.social and @davidandacat.bsky.social

www.nature.com/articles/s41...

Paper just accepted

Gibson et al. "On the stability of gradient descent with second order dynamics for time-varying cost functions" Transactions Machine Learning Research (2025)

openreview.net/forum?id=Hlz...

SAMPL-seq reveals micron-scale spatial hubs in the human gut microbiome - Nature Microbiology Split-And-pool Metagenomic Plot-sampling sequencing (SAMPL-seq) can be applied to complex microbial communities to reveal spatial co-localization of microbes at the micron scale.

Happy to share our new paper in @naturemicrobiol.bsky.social on mapping spatial relationships of the human gut microbiome. We identified distinct spatial hubs between gut bacteria that reflect sub-community assemblies at the micron-scale. Led by Miles Richardson & co.
www.nature.com/articles/s41...

Ethics of Caring - National Nursing Ethics Conference on LinkedIn: #nursing #ethics #nursingethics #clinicalethics #socialworkers #socialwork… “We can’t underestimate the power of storytelling…stories have a way of speaking to the humanity within each of us and can help remind us of the humanity in…

Join us in LA this March!

Are regret based adaptive control strategies deployable?
arxiv.org/abs/2501.04572

How can I run two isolated installations of Homebrew? I want to install and run two versions of Homebrew simultaneously on an Apple Silicon Mac: an ARM64 version, and an Intel version running under Rosetta 2. I know I can prepend any brew command with...

This post has saved hours when trying to get old x86 bioinformatics pipelines working on Mac arm64 (like pplacer)

stackoverflow.com/questions/64...

It’s easy to think this is about politics, but Ran’s actually just subtweeting my working style

As further evidence you should send them the @abrahamgihawi.bsky.social paper that invalidated and lead to the retraction of the Nature paper about a cancer microbiome.

at this point though we are speculating because there doesn't seem to be enough information in the paper or code for us to really see how the inference was actually performed

you can have leave one-out testing that respects the test train split... you just perform n rounds of train on (n-1) test on 1

This kind of mistake happens way too often