Back

A nutrient-derived dietary gradient links gut microbiome composition, network structure, and host physiology

Xu, J.; Riabchenko, M.; Bartolomaeus, T. U. P.; Mueller, A.; Hannibal, L.; Huber, R.; Jeitler, M.; Jarquin-Diaz, V. H.; Forslund-Startceva, S. K.; Storz, M. A.

2026-07-24 microbiology
10.64898/2026.07.24.740531 bioRxiv
Show abstract

Diet is an important determinant of gut microbiome composition and host physiology. However, dietary exposure is often represented using categorical groups that may not fully capture variation in nutrient intake or its associations with gut microbiome and host biomarkers. Here, we analyzed diet-microbiome-host physiology relationships in healthy adults with long-term adherence ([≥]2 years) to omnivore, vegetarian, or vegan diets. We integrated four-day weighed dietary records, blood biomarkers, and 16S rRNA gene sequencing to characterize dietary intake, host physiology, and gut microbiome composition. We derived nutrient intake gradients by dimensionality reduction of representative nutrients and identified the second nutrient intake gradient (NIG2) as a quantitative axis reflecting variation between animal- and plant-leaning nutrient profiles, broadly recapitulating categorical dietary groups. While NIG2 showed no association with microbiome within-sample diversity, it explained gut microbiome compositional variation that was not detected using categorical dietary groups. Mediation analyses identified nominal candidate microbiome-mediated diet-biomarker relationships involving thyroid-related, hematological, and vitamin-related biomarkers. Network analyses of NIG2-defined animal-leaning and plant-leaning subsets identified subset-specific microbial associations and differences in microbial community structure and potential interactions. Overall, this continuous nutrient-derived dietary gradient provides a quantitative exposure metric for characterizing diet-microbiome-host physiology relationships. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=97 SRC="FIGDIR/small/740531v1_ufig1.gif" ALT="Figure 1"> View larger version (23K): org.highwire.dtl.DTLVardef@104a825org.highwire.dtl.DTLVardef@bb2b33org.highwire.dtl.DTLVardef@114171org.highwire.dtl.DTLVardef@df9c38_HPS_FORMAT_FIGEXP M_FIG C_FIG

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.