My lab has been working on bacterial 16S-based microbiome analysis in mouse fecal samples in these two years. Recently I've been thinking to move on to Illumina shotgun metagenomic sequencing if we could afford. So I'd like to collect more information.
My goal is straightforward. I have a knock-out mouse with a special gut microbiota-associated phenotype. So I'd like to know the difference (bacterial species with abundance information) of fecal microbiota between the wild-type mice and my knock-out mice.
With MiSeq we found that our samples generally contain 50~60 different bacteria at the genus level.
So is de novo assemble required for shotgun metagenomic sequencing in our case?
How to start in terms of NGS coverage, depth and mouse numbers for shotgun metagenomic sequencing?
Thanks for your time.
Hi,
Shotgun metagenomic sequencing can give you more robust data about bacteria present in the gut of your model organism under your experimental conditions. If you expect to have 50-60 genus colonizers, this means that approx 500-600 species could be present in your model organism. With a averaged genome size of 4Mb this roughly means that the metagenome of your mice is composed by 2-2.4Gb. One tipical output from one lane of HiSeq2500 in 2x125 configuration has ~30Gb (15Gb in SE = ~120M reads), so you will have approx a 15X coverage. I recommend you mix 2 or maximum 3 samples per lane (always in the same run) under this conditions to get anough data to reconstruct the metagenome present in your animals. Don't forget to include at least triplicate of each treatment to permit statistical comparison considering the biological variance under same technical issues. With the raw data generated with this configuration you will be able to assembly a good metagenome with thousand of ORFs to map.
Good luck.
Alfonso
Thanks a lot, Alfonso. Your suggestion is very helpful. I am a bit confused with your calculation. So you said one lane of HiSeq2500 in 2x125 configuration has ~30Gb and if I put 3 samples per lane, I will get ~10Gb per sample. It is ~5x coverage. Will that be enough for metagenome ? Is de novo assemble necessary? Will I be able to get the bacterial transcriptome with that 5x coverage? Is there a good instruction to start for the shotgun metagenome analysis? Much appreciated!
Average 15X coverage, but far more than that for the dominant members of the community and far less (including under 1X) for the rare members. You will get assembly of the dominant players. Not a bad thing, but don't expect to assemble 500-600 genomes.
Assembly is not necessary, and the trend nowadays is not doing it, unless you want to recover genomes from the metagenomic data, e.g. new taxa. You can map taxonomy and pathways directly. Take a look to HUMAnN and other tools from bioBakery
http://huttenhower.sph.harvard.edu/humann2/
https://bitbucket.org/biobakery/biobakery/wiki/Home
Thanks for inputs. I knew it will be difficult to find the difference between rare members. I did not expect to find new taxa with my budget either. So if assembly is not necessary to identify the major differential species between two conditions, I'd go with 10Gb per mouse fecal samples. Probably 4 mice per condition in case one of them is really an outlier. Does this sound reasonable?
Typically in corhort studies already published, they obtained 4~8 GB per sample. I believe this may also be sufficient for mouse. If the composition is the only thing you are interested, you may not have to do assembly. Many tools can do taxonomic identification based on not only 16S rRNA genes, e.g. metaphlan. And you can also do some clustering work based on, like co-abundance correlations, using sparcc.
In practise, you could not expect to get many draft genomes from shotgun sequencing, except the most dominant ones.
I would recommend doing metagenome assemblies if you want to understand the metabolic potential of your microbiome. In your case with the special phenotype that could be worthwile. Using assembly independent pipelines, like Xabier is suggesting, will be usefull for well characterized ecosystems, but can be biased. Assembly can identify pathways not well described in other databases.
All the comments are helpful. Thanks. We already sent out 2 samples for preliminary test with 10Gb per mouse fecal sample (Hiseq 2500). I will use that as a guide for following analysis.
- Badges
- Science method
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis
- genetic techniques
- Sequencing