Xiaowen Feng

[Xiaowen Feng @0xfxfxf]

@katharina_hoff @miamynta @lh3lh3 We used SILVA rRNA database and barrnap, both seem to support 18S and several other rRNA genes. I have only read about and tried 16S. The OTU clustering method was independent to rRNA type, although the cutoffs (species or genus boundary) was based on 16S literature.

[Xiaowen Feng @0xfxfxf]

Abundant species in a metagenome sample are not guaranteed to be easier to assemble. In our @lh3lh3 new preprint (https://t.co/5BdoEzC28o ), we proposed k-mer based and 16S rRNA based methods to measure metagenome assembly completeness. We also proposed a new algorithm (1/2)

[Xiaowen Feng @0xfxfxf]

@rozovr One issue I have is there is no checkM equivalent for plasmids. For bacteria MAGs, rescue heuristics and evaluations ultimately rely on checkM and rRNA evidence to make sense of, but for plasmid contigs I find it hard to say anything. Mock datasets are not too useful for both...

[Xiaowen Feng @0xfxfxf]

@humanearthling_ Please feel free to open issues on github if you have any problem with hifiasm-meta (or on other stuff I have over there).

[Xiaowen Feng @0xfxfxf]

It's been a while but hifiasm-meta from our lab with
@ChengChhy @DPortik @lh3lh3 is now online at Nature Methods, readcube link: https://t.co/tEhpAiHDus . We look forward to seeing more Hifi metagenome libraries for method development and biological insights.

[Xiaowen Feng @0xfxfxf]

@Amanda_Stahlke @Magdoll @phototrophic @PacBio I don't know, and depends on how much contamination are there. I think hifiasm-meta is almost compatible with hifiasm bin files. If you're interested in quickly trying it after a hifiasm run, I can push a minor change to make it fully compatible (with some disadvantages).

[Xiaowen Feng @0xfxfxf]

@drashwinkelkar @arcane_mahesh @lh3lh3 @PacBio You can try that, but it did not work well on a Q20 dataset. I'm not sure whether tweaking parameters would help.

[Xiaowen Feng @0xfxfxf]

Our lab @lh3lh3 developed a metagenome assembler, hifiasm-meta (https://t.co/iEFeLhK3xn), which reconstructs @PacBio Hifi libraries to tens to hundreds of complete circular bacterial genomes. We hope these high-quality contigs to offer a new means to study microbial communities.

[Xiaowen Feng @0xfxfxf]

We would like to thank @ChengChhy @DPortik for their help and suggestions throughout the course of hifiasm-meta's developement, and @ Wei Fan for sharing the chicken gut microbiome dataset with us. Thank you so much!

[Xiaowen Feng @0xfxfxf]

The assembly needs no manual intervention or auxilliary data. Sequence divergence between all pairs of the complete circular contigs is >1%, suggesting little redundancy. We are always looking to test on more samples and improve. Consider give it a try: https://t.co/H1jusjgw3m

[Xiaowen Feng @0xfxfxf]

@BioMickWatson @lh3lh3 @PacBio Here is the sheep dataset's assembly graph: https://t.co/vEDe6LJj2a . Thanks!

[Xiaowen Feng @0xfxfxf]

@tendersombrero @lh3lh3 @infoecho @PacBio I agree with Heng. If the sample has very high coverage and you'd like to play with hifiasm-meta anyway, maybe try hifiasm-meta with --force-preovec --lowq-10 50 for the read selection, and then feed the bin files to vanilla hifiasm for final graphs.

[Heng Li @lh3lh3]

We are looking for metagenome HiFi datasets as each metagenome sample is unique. We need to learn more. If you have data and are happy to share, please let @0xfxfxf and me know. Thank you.

[Heng Li @lh3lh3]

Xiaowen Feng @0xfxfxf developed hifiasm-meta, a fork of hifiasm specialized for metagenome assembly with @PacBio HiFi reads. It can produce ~100 circular genomes from one sheep gut sample. Have a try: https://t.co/sMj70D77E5