A real beginner's question on metagenomics! Plz help!

[metaStudent]

Hi, I'm very much new to the field of metagenomics and sequencing. I have been trying to find out answers for this but I haven't been too successful probably because the answer is too obvious or my understanding is completely plainly wrong... Anyway, it would be great if some of you could enlighten me with an answer!

In metagenomics, say a soil sample is taken for sequencing. Now, fragmented genomes from this mixture of species are sequenced say by 454 sequencing. My question is, will there always be some overlapping regions to guide de novo assembly? If I am understanding this correctly (probably not!), there will be abundant species in the sample and their genomes will be fairly easy to assemble as there are lots of them, but for the less abundant species, there won't be enough of them to have enough overlapping regions of the genomes for de novo assembly. I know this is a bit extreme, but what if there happens to be just one member of a novel species in the sample? is there just no way for the whole genome of this species to be sequenced?

[rskr]

Maybe try assembling a couple polyploids with samples from various different strains of the same species, heck try assembling a single species from a single sample accurately, then figure whatever is in the meta soup is pretty close to nonsense alpha=.0001, especially after you take into account 454 is only accurate in high coverage regions and you probably only have at most a couple of 454 runs. So you might as well give up, or at least make up some results.

[ECO]

Maybe try assembling a couple polyploids with samples from various different strains of the same species, heck try assembling a single species from a single sample accurately, then figure whatever is in the meta soup is pretty close to nonsense alpha=.0001, especially after you take into account 454 is only accurate in high coverage regions and you probably only have at most a couple of 454 runs. So you might as well give up, or at least make up some results.

While this could've been said nicer (), I mostly agree. New species discovery by soil shotgun de novo sequencing is a hard problem.

Maybe with a few billion long reads....but i doubt with current technologies/throughputs. Either that or single cell sort + WGA + WGS.

[metaStudent]

Thank you, but I'm sort of more confused...

My question was, will there always be some overlapping regions to guide de novo assembly even for those species which are less abundant in the sample?

Really sorry if I failed to understand your previous answers.

[ECO]

There will never always be anything in biology...

Your question is not really answerable with real facts, unless you already know how many species are present in your sample and at what abundances. It's a pretty easy calculation to estimate: how many species do you have, what are their relative abundances, how big are their genomes, how big is the insert size of your library, and how many reads will you generate. Of course there are more difficult questions of species similarity, repeat content, but the above is probably a good exercise for you to work through.

The conservative answer is (especially with a few million reads from a few 454 runs)...that you will not have enough coverage to de novo assembly an entire RARE organism out of a metagenomic pool, let alone be confident that any apparent overlap is biologically real.

[metaStudent]

Thank you ever so much! I think all the confusion started when someone told me it's possible to conduct whole genome sequencing from a metagenomic soup! I suppose de novo sequencing in metagenomics is a dream that this stage! Thank you for your answers!