Unsupervised binning of metagenomic datasets using cluster size insensitive fuzzy c-means method

摘要

Metagenomics uses random shotgun sequencing technology to sequence the genomes from uncultured microbial communities. Metagenomic dataset is a kind of unbalanced dataset, as it has a characteristic of uneven species abundance ratio. Fuzzy c-means method (FCM) is a famous clustering method but has a poor performance for unbalanced dataset. This paper introduces a cluster size insensitive fuzzy c-means method (csiFCM) to binning of mategenomic datasets with uneven abundance ratio. Experiment results on synthetic datasets illustrate that csiFCM is an effective binning method for metagenomic dataset and has an overall better performance than AbundanceBin and MetaCluster3.0.