Host trait prediction from human microbiome data for Colorectal Cancer

Microbiomes, micro organisms living in a host environment, play significant role in the regular activities and abnormalities of the host. Researches throughout the world has found symbiotic relations between the human microbiomes and human physiology, immunity, diseases etc. Outstanding development in sequencing technology has paved the way to analyze large number of samples with reasonable cost. Applying supervised machine learning techniques on these data can help to find out the most important microbiomes residing in the host environment and build a predictive model to classify unknown samples. In this study, we have applied different supervised classification algorithms a long with some e nsemble techniques to find a better predictive model to predict the trait o f human host for the prognosis of colorectal cancer. Our study finds that, tree based classification algorithms w orks best f or classifying the human microbiome data for colorectal cancer which are large, sparse and dispersed in nature. We have also identified important microbiomes that acts as a deciding factor behind colorectal cancer.