Hi Joseph,

Chemical removal of host DNA might also deplete the metagenomic DNA content. Assuming the DNA elimination to be uniform, the overall variance in the dataset may remain same, but you may lose strength of the signal.

For bioinformatics based sequence removal may not be full proof, but it's hassle free. You can use bmtagger, bmfilter, srprism programs by NCBI, with the amphibian host genome as database, and it will return ids of reads of host origin, which can be removed from the sequence file using a custom script.

I hope this helps.

Hi Joseph

I agree with Anubhav, chemical removal of host DNA might bias your final dataset. My personal experience is that several taxa disappears during host DNA depletion. Furthermore, it can be quite a bugger and not super efficient to minimise the host DNA. But if you wan't to do host depletion, then I would prefer fresh sample and a direct digestion of eukaryotic cell by differential lysing of eukaryotic and prokaryotic cells. And i would do metagenomics on both depleted and non-depleted DNA to recover most possible genomes.

This is of course a question of money, since sequencing metagenomics can be rather pricely.

Some considerations you could include for you discussion before starting up the project is:

• Have you investigated if this depletion is a necessity for amphibian gut? Like with 16S primers and qPCR?

• Also if the diversity is known already, if it is low, then i guess you could be less worried on losing data to host DNA, since you would have less genomes to resolve.

I hope this helps

Thanks for the feedback Anubhav. It does seem like host DNA reduction will cause its own issues, including non-uniform microbial DNA depletion (which isn't good for microbial ecology applications). If there are enough reads, dealing with this as a bioinformatics issue seems to make more sense from what I have been reading.

Hi Joseph,

I also agree with the last comments, bioinformatic host DNA reads removal is the best way with minimal downstream biases. I have not worked with amphibians, but in human fecal samples, the percentage of host DNA is usually low (less than 1-5% of the total reads). But it really depends on many factors, e.g. patients with GI problems and high inflammation, tend to have more host DNA in their samples. Regarding tools, you can use short-read aligners such as Bowtie2 or BWA to map your samples against the host genome/s, and removed such reads.

Good luck

Great, thank you for the feedback Jacob and Vadim. I will Be opting for no host reduction for this experiment. I may try adding a few samples that have had host DNA removed as a sort of side experiment/comparison. As this is a NovaSeq experiment, I unfortunately don’t want to add many more samples and another lane/run is over budget.