Advantage of direct sequencing, if you use a new generation sequencing system you can have your results in a really short time.

Disadvantage of direct sequencing, you have to obtain a really good quantity of the DNA for a representative result, and you have to perform the extraction protocol many times.

Advantage of library, you will stabilize your copies in a bacteria so you can make as many repetitions of your experiment as you want.

Disadvantage of library, a lot of work is needed, the time for the obtention of the results is long, imagine that you have to extract DNA and sequence every colony you´ve got

As far as I know, if you are talking about NGS, 16S primers (v2-v3 regions of 16S) amplify about 300-400 pb (if you are lucky) and have some biases. One advantage of the library is that you can get until "species" rank, but cloning is not 100% efficient and maybe you are losing some information. In NGS usually you go until "family" or "genus", but you have better rarefaction index.

Making a clone library (the old fashioned approach) will give you longer reads, because you can amplify (near) full length 16S genes and sequence them from both ends, so you will get 1400-1500 bps reads. This will allow you to do a very good phylogenetic/taxonomic assignments, even if you are exploring a habitat that has a lot of novel (previously unseen) sequences. But picking up the clones is a lot of manual work, and Sanger sequencing will cost a lot. I have picked up and sequenced 1000s of clones, but it was a lot of work. One will typically obtain between 100 and 500 clones per sample.

Pyrosequencing (454) will give you shorter reads (500 bps), but you will get 1 or 2 millions of reads. If you use barcoded primers, you can combine a couple of hundred samples per run, so you will still get 3000 - 7000 of reads per samples. So you will have a much deeper sequencing effort for a fraction of the price (compared to the clone library approach). The shorter reads are still good enough for taxonomic assignment up to genus or sometimes species level.

Illumina type of sequencing will give even more reads, but the reads are shorter (100 - 150 bps), so it is harder to assign your sequences to a taxonomic group.

So it is really a trade off between numbers of reads and ability to assign them to taxonomic groups.

I hope this helps!

Amplification, cloning and sequencing require a strong effort both in terms of work and of money because in this strategy you will perform Sanger sequencing. Advantage is that sequencing from both ends of the insert gives you very long sequences useful to have a correct assignment of the species to the correct taxonomic group.

Next gen sequencing requires less work and is cheaper; you can easily perform multiplexing (up to 100 samples per run or even more) and have a very detailed result of the abundance of the community microorganisms (OTUs). Disadvantage is read length (400-500 bases using FLX titanium sequencing) determining a more difficult assignment of the reads to the taxonomic groups.

Thanks everyone. This discussion was excellent and I could get more information over both strategies. One last question: using degenerated primers, is the direct sequencing applicable or considering the big range of the primers this approach could be inefficient?

See attached paper

Thank you Fransisco, but I think you forgot to attach the paper, or I just do not know where to find it on your answer. Could you check it again? All the best.

Direct sequencing would only work if you have a pure culture (e.g. of an unknown, cultured, organism). However, if you are working with mixed bacterial communities (e.g. soil, or feces), it won't work because you will have a mixed read. The first couple of nucleotides might still be readable, because you will start sequencing from a conserved region, but after that the reaction becomes mixed (all nucleotides will be incorporated), so you cannot interpret the results.

Although this one is about real metagenomics not just 16S. Sorry

Incidentally I think doing only 16S by NGS is largely a waste of time and money. For a very low price you can do 16S and other genes that sometimes are much more interesting and informative.

I do agree with Elizabeth in that direct sequencing could prove troublesome because the sample is mixed and the readings wont be clear enough to determine the sequence. Is there a way you can sequence individual organisms by NGS without having to clone in bacteria?

A huge advantage of cloning is that you create a DNA database that you can draw upon for doing replicates. It's a lot of work though, and occupies a lot of space, depending on the size of your samples.

Francisco, could you give more details about which genes you think would be more informative to do together with 16S? I'm thinking of doing roughly the same strategy for determining diversity of areas for photossynthetic Eukaria, using nuclear ITS and plastid sequences.

Hello Julio and everyone (tudo bem?),

It is really not a problem to sequence a complex amplicon mixture. The sequencing procedure takes care of separating single molecules from the mixed PCR amplicon, which will then be individually sequenced.

It is a very good approach for work aiming to survey microbial populations in complex community samples (stool, soil, water, sediment etc). No cloning or separation of the initial amplification mixture is necessary and the technology is perfectly comparable to Sanger based surveys (there is lots of literature on this, Rob Knight and other have published extensively on this).

NGS has the same caveats of working with a mixed population of templates as with traditional approaches (DNA extract -> PCR-> cloning -> pick clones/minipreps -> sequence): the extraction and PCR are essentially the same. You bypass the cloning and clone picking steps and move directly onto the sequencing, but the initial biases of extraction and PCR are about the same (you can actually decrease those because it becomes feasible to do repeated DNA extractions and with replicate PCR amplifications).

And the number of sequences obtained per sample is far greater than you'd get from cloning approaches. You can always go back to your PCR product to sequence more if you wish, or re-amplify and sequence again, or as I said: barcode several PCR reactions individually and sequence them all at one (same applies to multiple samples).

The advantages for survey work are numerous and when you put cost and time, the amount of the sampling you can do, as well as sequences per sample you get, the cloning based approaches becomes quite unattractive.

One alternative would be to use Illumina sequencing to get metagenomes. You can still obtain the taxa survey, plus several other gene sequences that might be of interest.

The Huttenhower lab has put out an interesting tool (metaphlan) which give you taxa derived from metagenomic sequences and used several genes across the different taxa to help identify the community composition. It's a new tool but seems promising.

best,

Chris

ps: we work with degenerate primers all the time in several types of samples (blood, stool, BAL..., for bacterial, viral as well as human targets, starting with DNA or RNA). The PCR reaction is a bit more trick to get clean, and it will off course be dependent on the sample/amplicon at hand (nothing that a bit of optimization won't cure), but that would be expected anyway.

To add to the comments above, the NGS approach for 16S-based taxonomic assignments is to date the most effective approach available to date in terms of workload, time, money and quantity of data. Short read technique such as Illumina are useless, and only long-read accurate techniques such as 454 can do a proper job.

The disadvantages linked to the length of the sequenced region are in fact a great improvement over the numerous biases introduced by cloning and the workload generated.

Last but not least, the large metagenomic datasets are handled by pipeline which makes the analysis on a million reads almost easier than that of a 100 a couple years bacK.

In case some very interesting bug show up in the data, it is always time to go and fetch it using the classical cloning approach if you want the complete 16S sequence. In that way, 454 helps to focus the sequencing on the real novelty.

You have to be aware that most of the drawbacks associated with NGS for taxonomic assignment, i.e. PCR bias, gDNA extraction bias, etc. are also applicable to the cloning/sequencing approach, and that on top of that this later will had a further limitation due to cloning bias, clone selection and limited number of sequences.

Last, if biases are a concern, it is easier, and recommended to perform for the same sample at least duplicate sequencing of a sample; In many cases, we also do amplification of the target population with different primer pairs for the same sample, to compensate for the unavoidable primer biases.

You may find your answer in below publication

Article Techniques used to characterize the gut microbiota: A guide ...