nf-core/taxprofiler

Decontamination: add hostile for host removal

Opened this issue · 5 comments

jfy133 commented

Description of feature

Add the HOSTILE tool for host decontamination

https://www.biorxiv.org/content/10.1101/2023.07.04.547735v1

https://github.com/bede/hostile

It's interesting to see the various reference genome options available in HOSTILE, such as combining the human T2T genome with HLA, argos985 and mycob140. This combination can filter out human reads from the MTC region and retain more microbial reads. For additional details, please check: bede/hostile#27

I think the reference genome is often the most important factor!

so indeed this will be helpful

@jfy133 I agree with you.
In our clinical cases, it's quite often to see that false positives caused by human rRNA.
Therefore I am considering to include human rRNA databases in the reference genome. Even though the human Telomere-to-Telomere (T2T) assembly offers better results compared to GRCh38, assembling rRNA remains a challenge due to its high repetitiveness and significant variations between individuals, ages, and even certain types of cancers.

Not a bad idea! But I'm not sure if at that point it is worth investing so much into the host reference, rather than checking whether the microbial reference genomes are clean...