The authors collected 11 colorectal tumor samples and applied three technologies to detect the microbiomes in these samples, including 16s rRNA sequencing, RNA-seq and MinION sequencing. MinION sequencing is a platform that can produce long sequencing reads. The results in the manuscript illustrate the advantages of MinION sequencing in detecting microbiomes. These advantages include MinION requires lower sequencing depth, is more sensitive, and is more discriminative than other methods.

1. Did authors summarize the quality of fastq files? Some software can do this analysis, such as FASTQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/).

2. It will be great if authors could provide the details and parameter settings for the software that were used in the study.

3. What are the percentages of level of concordance reported in Figure 1? Are they the coefficients of Spearman’s correlation? Is it possible to report p-values to show whether the concordances are significant or not? It will be great if authors can provide the details of the Spearman's correlation analysis.

4. Also for results in Figure 2 and 3, is it possible to report p-values to show that MinION is significantly more concordant to RNA-seq than to 16s rRNA?

5. Based on the results in Figure 1-3, authors conclude that MinION has more concordance to RNAseq, which is reasonable. But how to make the conclusion that MinION and RNA-seq are superior to 16s rRNA sequencing for taxonomic classification? Since we don’t know the ground truth of which microorganisms exist in the CRC tumor samples.

6. Is it possible to do experiments to valid the results reported in the manuscript? For example, in the section “Detection of colorectal cancer-associated bacterial species”, authors reported that MinION read can cover a TetR transcriptional regulator that involved in antibiotic resistance and an amino-histidine dipeptidase, but RNA-seq reads can’t. Is it possible to valid this result by experiments such as PCR?