Fig. 11

DA analysis of stool microbiomes of colorectal cancer. a Venn diagrams of DA species across four CRC datasets. Shown are sets detected by t-test, DEseq2, edgeR, metagenomeSeq, and t-test applied to imputed/denoised data (mbImpute, SAVER, and mbDenoise-zinb-cov). We used a FDR threshold of 0.05. b Percent stacked barplot of CRC-enriched species identified across four datasets by each method. Condition 1-4 represent species identified in one, two, three, and four datasets. Numbers marked red correspond to the counts found in each condition. c Heatmap of − log10(FDR-corrected P values) of the 23 common species identified by mbDenoise-zinb-cov across four datasets. d The frequencies of seven CRC biomarkers discovered across four datasets in each method. Condition 4-2 represent biomarkers identified in four, three, and two datasets