16S metagenome sequencing directions(F/R)

[jiaco]

Dear SEQanswers

I have a very general question and would love to have input from at least a handful of different people with experience in this field.

We are doing 16S profiling of bacteria from human/mouse host tissue samples and I find this pdf on the net that seems to be a decent starting protocol that is in use worldwide for exactly this type of study:

http://www.hmpdacc.org/doc/HMP_MDG_454_16S_Protocol_V4_2_102109.pdf

In the protocol, the barcode is added to the R (reverse) primer only and sequencing is done in a single direction using that R primer.

Strangely, the local sequencing center claims that they had problems with single direction sequencing and they only sequence in both directions, but not paired. This results in barcodes on both primers and compatibility issues with mothur/qiime pipelines for processing flowgrams. They even went as far to say that the new Roche kits do not do single direction sequencing.

Sadly, I do not have the experience nor the knowledge to support or refute such claims. But I am hoping that people here do and can help me arm myself for my next meeting so that I can do 454 16S sequencing in a standardize manner that is most compatible with the software that I want to use to process the output.

[vamosia]

As far as know 454 does not provide unidirectional amplicon sequencing. So for now you're stuck with sequencing both direction. (You can theoretically use the Lib-L shotgun protocol to perform a unidirectional sequencing, but this protocol is shogun protocol not optimize for amplicon

Alternatively, you can go through you fasta files and only used the reverse reads in your analysis completely ignoring the forward reads.

[ajthomas]

There is an application note on unidirectional sequencing of amplicons using the Lib-L kit. I've done it a number of times and it works just fine. The one thing I do, however, that seems to help, particularly with homopolymer stretches, is I reduce the amount of amplification primer to about half of what the protocol calls for. Some suggest reducing it to 1/4 of recommended amount, but I haven't tried that.

[Nicole 454 Sequencing]

Hi Jiaco,

There are a couple methods you can employ for unidirectional sequencing. As Ajthomas pointed out, there is an Application Note (App no. 001-2009) located on the customer accessible area of www.my454.com. This method uses the Lib-L emPCR kit for unidirectional amplicon sequencing.

Unidirectional sequencing is possible with the Lib-A emPCR kits by processing only the A (Forward) or B (Reverse) set of reagents. The drawback to this method is half of the emPCR kit is wasted.

Nicole

Technical Support Scientist
454 Life Sciences, A Roche Company

[cliffbeall]

Jiaco-

We have been using the HMP primers in that link that you give. They are designed to be used with Lib-L reagents. We have had some problems with low numbers of sequences passing filter, probably what your sequencing people are referring to. We got some improvement by lowering the amount of primer and copies per bead in the emulsion PCR. In fact, we reduced copies per bead from 2 to 0.25, and its not clear we've hit the optimum yet for that. More discussion here:

http://seqanswers.com/forums/showthread.php?t=11612

Best, Cliff

[jiaco]

Thanks to all that have replied so far. It looks like I am going to have to learn a lot more about the technology aspects of this before I can really understand.

But 2 simple questions:

1) Is the HMP protocol in use worldwide? I was under the impression that if one were to do 16S studies and wanted to compare different analyses, that this protocol was optimized for that and would be the best on to use.

2) If we use the HMP primers and just keep reads that originate from the barcoded primer, the other reads that we throw out would be in fact sequences from the same amplicon but just in the opposite direction, right? We would not really be loosing data. I really cannot get my head around sequencing in both directions that is not paired.

Thanks again to all and to cliffbeall (honestly I might have forgotten to search first).

[cliffbeall]

Jiaco--

I think quite a few people are using those primers, at the least the HMP has generated a massive amount of data with them (300 subjects, 18 body sites). They are designed for sequencing with the LibL reagents, though. At the time the protocol was developed that was the only choice.

If you did want to do sequencing with the LibA reagents you would need to redesign the 454 adapter part of those primers. It might be worth doing, even with the optimization we have done, we're only getting about 40% keypass reads through the quality filters.

You could still do unidirectional with LibA as Nicole mentioned.

If you did want to stick with the HMP protocol I could put your sequencing core in contact with ours. Drop me a PM or email if you want to try to set that up.

[jiaco]

Just got off the phone with a company that provides 454 and have learned a bunch of stuff. Would love to get feedback if anyone has anything to add.

Basically, they agreed that double stranded sequencing is not beneficial for 16S and use the Lib-L shotgun.

They also said that they have problems with XL+ that are still not resolved and propose to use Titanium as they have had success with that.

They were also aware of the issues cliffbeall pointed out about primers and copies per bead and it sounds like they have optimized on that end.

In addition, they have seen the 100bp reads that resemble primer dimers and have modified their PCR cleanup to remove those.

However, they want PCR products under 500bp and suggest we use V3-V4, which are not HMP primers.

A final point is that they use Roche barcodes and have never heard of Hamming and Golay.

Overall, I was reassured on many levels by this new company, versus the previous local core facility, but wonder if I need to widen my search for a 454 provider.

[tokikake]

Basically, they agreed that double stranded sequencing is not beneficial for 16S and use the Lib-L shotgun.

We do a lot of 16S sequencing and only do unidirectional sequencing with the Lib-L Kit.

They also said that they have problems with XL+ that are still not resolved and propose to use Titanium as they have had success with that.

We also had problems after our upgrade to XL+. We compared the same libraries with Titanium and XL+ (both Lib-A and Lib-L) and could not see a clear difference between them. Finally our XL+ had to be fixed and now I am satisfied with the performance. But the amount of reads and the length varied and depend on the origin of the samples.

In addition, they have seen the 100bp reads that resemble primer dimers and have modified their PCR cleanup to remove those.

I think every lab dealing with 454 is aware of this problem; in our case we encourage the people to do gelextraction and then bring us the purified product. Without that, we won't sequence it

However, they want PCR products under 500bp and suggest we use V3-V4, which are not HMP primers.

We have good results with the primer pair you mentioned. Now we switch to the V3-V5 region, to benefit from the longer read length of the XL+.

A final point is that they use Roche barcodes and have never heard of Hamming and Golay.

We also use the MIDs from roche; it's easier and we don't have to adapt the parse file after every run.

[cliffbeall]

Tokikake-

What percentage are you getting in terms of keypass reads passing all filters for the HMP V1-V3 or V3-V5 primers and Titanium?

We haven't been doing gel purification, but have done 2X ampure (calibrated for the size cut-off as in the Roche small fragment removal protocol). It wouldn't be feasible to do a gel until after the pools had been made, but it might be worth doing if we can get more sequence per run.

[HMorrison]

Thanks to all that have replied so far. It looks like I am going to have to learn a lot more about the technology aspects of this before I can really understand.

But 2 simple questions:

1) Is the HMP protocol in use worldwide? I was under the impression that if one were to do 16S studies and wanted to compare different analyses, that this protocol was optimized for that and would be the best on to use.

2) If we use the HMP primers and just keep reads that originate from the barcoded primer, the other reads that we throw out would be in fact sequences from the same amplicon but just in the opposite direction, right? We would not really be loosing data. I really cannot get my head around sequencing in both directions that is not paired.

Thanks again to all and to cliffbeall (honestly I might have forgotten to search first).

We pioneered 16s tag sequencing on 454 back when it was a GS20 (Sogin et al., PNAS 2006). We are involved in an HMP demonstration project and resisted the sequencing centers' push to use LibL and shotgun processing for tag sequencing. Up until just recently (see other thread today), we were getting 800K reads of 535 nt average using the LibA kit and amplicon/rCAFIE-FLX processing pipeline. We sequence from one end of the amplicon only (640-650 bp amplicon) and since the LibA kit comes with both A and B reagents, we had to invest in primers that had either A or B adapters at the end from which we want to sequence.

We did a side by side comparison of yield and quality with data from the centers in UH2 phase of the project and were way ahead. Unfortunately, their SOP was already in use. They also use a single primer pair that misses phylogenetic groups, but that's another issue.

If you sequence using both A and B reagents and just keep the barcoded reads, you are throwing out half the data. Bidirectional sequencing is not mate-paired sequencing.

We are planning to upgrade to the FLX+, but not necessarily for amplicons.

[Carbon Copy]

jiaco,

These guys used same universal bacterial primers 24F and 534R for 454 nextgen metagenomic sequencing. The MIDs were incorporated into forward 24F chimeric primer, not 534R reverse primer like in HMP document.

http://www.ncbi.nlm.nih.gov/pubmed/20581188

In their work sequences were trimmed resulting in an average sequence length of 230 bp. Therefore a part of 16s rRNA gene which has been used for phylogenetic analysis was V1 + V2 regions. Apparently V3 region was ignored / lost due to sequence trimming. If they followed HMP protocol and attached MIDs to the 534R reverse primer they would have sequenced V3 region but lost V1 and V2. Please, correct me if I am wrong.

Different 16s variable regions are more important for identification of different groups of microorganisms according to this publication:

http://www.ncbi.nlm.nih.gov/pubmed/17391789

I assume that longer reads allow much more accurate speciation. That is why use of the Titanium 454 chemistry is more important than sequencing direction. Again it’s only my assumption since I have never done it. I am in the exactly same situation as you are: in the process of figuring out how to use our new 454 Junior for bacterial metagenomic studies.

In fact what worries me the most is biased or unequal amplification which may generate species abundance picture totally different from the reality. For instance in the first of following publications authors have shown that amplification involving 24F 16s rRNA primer can be very biased towards certain bacterial species.

http://www.ncbi.nlm.nih.gov/pubmed?term=18296538

http://www.ncbi.nlm.nih.gov/pubmed?term=17343679

http://www.ncbi.nlm.nih.gov/pubmed?term=8593063

I wonder if this problem has ever been discussed here or anywhere else by nextgen community.

[jiaco]

@HMorrison - we actually barcoded both ends. I have been working the last week or so getting programs in place to basically prove the data we have is crap and that we deserve a refund from the sequencing center.

@CC - Lots of good reading, but until I can actually get some decent data using a decent sequencing strategy, I have no clue about the issues in those papers.

We are about to use a private company to do 454 and in parallel run a phylochip analysis at second genome. I have high hopes that this combination of approaches can give us some answers.

Really glad that this forum exists, the help and information I have gotten here has been great.