Fig. 7
From: VirFinder: a novel k-mer based tool for identifying viral sequences from assembled metagenomic data
Two-way hierarchical clustering of viral contig bins and human gut metagenomic samples from a liver cirrhosis study [36]. Rows are contig bins (n = 116) comprised of 2657 viral contigs predicted by VirFinder, and columns are samples from healthy (blue) and cirrhosis (red) patients (n = 80 and 76, respectively). The heatmap depicts bin coverage across samples calculated as the averaged Reads Per Kilobase per Millions mapped reads (RPKM) of contigs in each bin. Bins form three coherent clusters that generally correspond to bins that are found in nearly all (95% on average), most (82% on average), or some (33% on average) samples. With the exception of two outliers (in grey), samples belong to three major clusters, A, B, and C, and clusters A and B each have two coherent subgroups (A1, A2, B1, B2). Stars depict bins that are positively (red) or negatively (yellow) associated with cirrhosis samples chosen using the lasso method for subset variable selection