Best Practice To Find Human Contamination In Microbiome?

Question

[Iddo]

I have gut human microbiome data, I would like to filter out human sequences. What would be the best practice? I thought of megablasting the reads (454) against the human genome and removing 100% identities. Any ideas to make this better / shorter?

Thanks,

Iddo

Answer 1

[lh3]

BWA-SW if you really care about speed. Or ssaha2 or blat if you are more comfortable with traditional tools.

Answer 2

[Casey Bergman]

If you have shotgun metagenomic data, this sounds like a reasonable plan - though you could swap out megabalst for a different/faster aligner like BLAT, and allow some degree of mismatch for sequencing error. If you have have targeted metagenomic data (e.g. a 16S amplicon), then you could additionally impose a filter for contamination hits to be off-target (e.g. non-16S).

Answer 3

[Wolfgang]

If BLASTN against the NCBI-NT is possible for the reads, you could also try to use CARMA3 to filter out human sequences...