Thanks! Totally! We have a preprint using a similar framework to look at mutational signature heterogeneity across tumors (collab with @nalcala.bsky.social doi.org/10.1101/2023...) and we also used this framework to look at heterogeneity of immune cell infiltration (analysis not published yet).

#mevosky #microsky

8/8 FAVA is the culmination of many conversations and collaborations. Thanks especially to my co-authors, @ksxue.bsky.social and Noah Rosenberg–I truly feel like I won the collaborator/mentor lottery. And thanks to you for reading! Paper is here if you want to read more: doi.org/10.1073/pnas...

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!

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.

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.

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).

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.

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: maikemorrison.github.io/FAVA/article...

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 variation would have been missed by stool sampling! We also find taxonomic variability >> functional variability, evidence for functional redundancy!

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:

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).

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.

Reading "The Science Bargain" by Rush Holt: "Democracy is at risk when it becomes simply a contest of opinions not grounded in evidence. When one opinion is as good as another–each asserted as strongly or even as deceptively as possible–democracy cannot survive. This is a call to science."