Decoding the language of microbiomes using word-embedding techniques, and applications in inflammatory bowel disease

Microbiomes are complex ecological systems that play crucial roles in understanding natural phenomena from human disease to climate change. Especially in human gut microbiome studies, where collecting clinical samples can be arduous, the number of taxa considered in any one study often exceeds the number of samples ten to one hundred-fold. This discrepancy decreases the power of studies to identify meaningful differences between samples, increases the likelihood of false positive results, and subsequently limits reproducibility. Despite the vast collections of microbiome data already available, biome-specific patterns of microbial structure are not currently leveraged to inform studies. Here, we derive microbiome-level properties by applying an embedding algorithm to quantify taxon co-occurrence patterns in over 18,000 samples from the American Gut Project (AGP) microbiome crowdsourcing effort. We then compare the predictive power of models trained using properties, normalized taxonomic count data, and another commonly used dimensionality reduction method, Principal Component Analysis in categorizing samples from individuals with inflammatory bowel disease (IBD) and healthy controls. We show that predictive models trained using property data are the most accurate, robust, and generalizable, and that property-based models can be trained on one dataset and deployed on another with positive results. Furthermore, we find that properties correlate significantly with known metabolic pathways. Using these properties, we are able to extract known and new bacterial metabolic pathways associated with inflammatory bowel disease across two completely independent studies. By providing a set of pre-trained embeddings, we allow any V4 16S amplicon study to apply the publicly informed properties to increase the statistical power, reproducibility, and generalizability of analysis.Author summary: The gut microbiome in humans has been implicated in a spectrum of diseases including inflammatory bowel disease, anxiety, depression, and Parkinson’s Disease, but thus far the associations between qualities of the gut microbiome and host symptoms are often not consistent across datasets. This may be because individual microbiome studies generally contain relatively small sample sizes, because some microbes are present in some populations but not others, and because microbial metabolism is dependent on the environmental context at hand. At the same time, there is a plethora of publicly accessible data describing the gut microbiome compositions of thousands of individuals in addition to their disease status, dietary habits, and lifestyle choices. We have employed a word embedding algorithm to map gut microbes from massive public datasets to vectors of real numbers which then represent relationships between microbes, or microbe-microbe co-occurrence patterns. We then use this mapping to learn more about what the gut microbiome of individuals with inflammatory bowel disease looks like, and find that mapping microbes to their vectors allows us to generalize results from one population to another more accurately.