Viral metagenomics sample with a high concentration of bacteria

[chariko]

Hi all,

I have analyzed a total of 7 samples coming from the bacteriophage fraction of human faeces. Purification included resuspension of the sample in buffer SM and homogenized sample was then filtered by low protein absorption filters. Afterwards samples were DNAse treated in order to eliminate non viral DNA or non capsided DNA. Also, samples were treated with chloroform to eliminate membrane vesicles containing DNA. Then, posterior membrane filters allowed to eliminate another residual bacteria.

Samples were sequenced in an Illumina Miseq paired end 2x150 bp. Samples were de novo assembled and taxonomically classified.

The problem is that a high proportion of bacterial DNA was detected in my samples so I wonder if someone could know what happened.

1) Could it be possible that bacterial DNA has been integrated in bacteriophages?

2) Should purification have included any other step?

Any other clue would be appreciated

[Richard Finney]

What kind of bacteria are you getting?
Fecal or other?

[chariko]

What kind of bacteria are you getting?
Fecal or other?

51% of Bacteroides vulgatus whose habitat is distal small intestine of humans.
10% of Bacteroides thetaiotaomicron a dominant member of our normal distal intestinal microbiota
9% Porphyromonadaceae
5% Bacteroides helcogenes
5% Alistipides
5% Prevotella
5% others

5% Viruses

Showing this percentages it seems the purification didn´t work at all but I wonderered this may be due to the adquisition of the phages from DNA of the bacteria... any suggestion is warmly wellcolmed

[colindaven]

Seems quite good to me.

In the past samples from human/plants only had viral loads way below 5%. I'd say about 0.1 % or less is more likely if you don't do purification.

There will always be huge contamination in whichever habitat you are trying to sequence viruses in, since they are low copy number with tiny genomes.

The bacterial DNA you sequenced might well be remnants i.e. non-living bacteria which just didn't get filtered out.

If you are just doing 2x150 MiSEq you can go to HiSeq 2x150 or 2x250 using future samples to get more data from the 5% viruses you have.

[chariko]

Seems quite good to me.

In the past samples from human/plants only had viral loads way below 5%. I'd say about 0.1 % or less is more likely if you don't do purification.

There will always be huge contamination in whichever habitat you are trying to sequence viruses in, since they are low copy number with tiny genomes.

The bacterial DNA you sequenced might well be remnants i.e. non-living bacteria which just didn't get filtered out.

If you are just doing 2x150 MiSEq you can go to HiSeq 2x150 or 2x250 using future samples to get more data from the 5% viruses you have.

thanks colindaven for your answer, do you have any reference in which this percentages are showed?

[adam.geber]

Can you describe your DNase treatment more thoroughly? Also, what size filters were you using? Viral-like particle enrichment is quite difficult and needs to be tailored to each sample type in order to get a high proportion of viral-origin reads.

How were you preparing libraries from your DNA extracts? Do you still have those samples? Also, what software are you using for taxonomic classification? Classification will depend entirely on what you're using as your reference -- can I ask what you're using for this as well?

Depending on your contig size you might be able to look for the colocalization of reads annotated as bacterial and viral within a single contig (although this could also be a sign of misassembly).