1/ Hey y'all, I'm excited to share my latest paper, which is out now in PNAS! We introduce FAVA, a statistical framework to measure compositional variability across microbiome samples. If you want to measure variability across a stacked bar plot, FAVA is for you! Paper: doi.org/10.1073/pnas... - ThreadSky

maikemorrison.bsky.social • 3 days ago

1/ Hey y'all, I'm excited to share my latest paper, which is out now in PNAS! We introduce FAVA, a statistical framework to measure compositional variability across microbiome samples. If you want to measure variability across a stacked bar plot, FAVA is for you! Paper: https://doi.org/10.1073/pnas.2413211122

Comments

vehuardo.bsky.social•3 days ago

Cool - This looks super-useful!

dnakendra.bsky.social•3 days ago

📌

kbird28.bsky.social•3 days ago

This looks really useful - thank you for this work!

maikemorrison.bsky.social•3 days ago

2/ FAVA is implemented in an R package available for download from CRAN (just run install.packages("FAVA") in R). For detailed guidance on the analysis of microbiome data with FAVA, check out the tutorial on the package website: https://maikemorrison.github.io/FAVA/articles/microbiome_tutorial.html

maikemorrison.bsky.social•3 days ago

3/ FAVA extends the pop-gen statistic FST (which traditionally measures variability of allele frequencies across multiple populations) to instead measure variability of taxonomic or functional abundances across multiple microbiome samples (Fig. 1). All it needs is an OTU table, or similar.

maikemorrison.bsky.social•3 days ago

4/ FAVA can analyze any relative abundance data, including taxonomic categories like OTUs and species, and functional categories like CAZymes and COGs. You can directly compare FAVA values computed on data with different numbers of categories (e.g., variability of bacterial species vs. families).

maikemorrison.bsky.social•3 days ago

5/ FAVA can account for phylogenetic similarity among taxa and can differentially weight samples, which is especially useful for time series data (Fig. 3). FAVA can be computed in sliding windows to see how variability changes over space or time (Fig. 4). We demonstrate FAVA in 2 examples.

rcthompson.bsky.social•3 days ago

This looks cool! Do you think FAVA could also be useful for e.g. looking at relative abundances of different immune cell types in blood samples (perhaps using branching differentiation paths to define similarity)?

maikemorrison.bsky.social•22 hours ago

Thanks! Totally! We have a preprint using a similar framework to look at mutational signature heterogeneity across tumors (collab with @nalcala.bsky.social https://doi.org/10.1101/2023.11.23.23298821) and we also used this framework to look at heterogeneity of immune cell infiltration (analysis not published yet).

maikemorrison.bsky.social•3 days ago

6/ (Ex. 1) In microbiome samples collected along the GI tracts of individuals from 7 ruminant species, FAVA peaks in the small intestine–so lots of variability would have been missed by stool sampling! We also find taxonomic variability >> functional variability–evidence for functional redundancy.

maikemorrison.bsky.social•3 days ago

7/ (Ex. 2) In longitudinal microbiome samples from healthy humans taking an antibiotic (collab with @ksxue.bsky.social), FAVA is much higher post-abx than pre-abx, and just half of the 22 subjects return to their baseline FAVA level (computed in 6-sample sliding window) during the 30 days post-abx!

urban-sci.bsky.social•3 days ago

Awesome. Will use!

Comments

Posting Rules

Reply