Huge difference in assembly stats between metaSPAdes and metaviralSPAdes when assembling gut viromes

[ilnamkang]

Hi,

I performed assembly of my gut viromes uisng both metaSPAdes and metaviralSPAdes.

It seems that the two pipelines give very different results.

Below is the summary stats of the contigs.fasta files I got.

Sample1	No. of contigs	Total length of contigs	mininum length	average length	maximum length
metaSPAdes	90,009	61,672,558	128	685.2	307,715
metaviralSPAdes	98	3,119,519	2,146	31,831.80	181,094

Sample2	No. of contigs	Total length of contigs	mininum length	average length	maximum length
metaSPAdes	96,334	72,584,866	128	753.5	168,431
metaviralSPAdes	100	3,042,634	1,225	30,426.30	103,608

There are huge differences, except for the maximum length of contigs.

Could there be huge differences like these between metaSPAdes and metaviralSPAdes?

My virome sequences were obtained from fecal samples through procedures including viral particle enrichment.

Taxonomic profiling using Phanta (https://www.biorxiv.org/content/10.1101/2022.08.05.502982v1, https://github.com/bhattlab/phanta) showed that ~90% of reads are from viruses.

Thanks.

[asl]

Tagging @Dmitry-Antipov

[Dmitry-Antipov]

Hi,
metaviralSPAdes was designed to extract viral part of regular metagenomes, so for viral enriched datasets I'd suggest to use metaspades.
Difference between metaviral and meta is expected, since metaviral does not output anything but some paths, suspcicious to be complete viruses (using coverage and graph topology), and metaviral outputs all the assembled sequence, including also fragmented viral genomes, bacterial chromosomes and whatever you sequenced.

[ilnamkang]

Thank you for your explanation.