Accuracy of a one way read

[emitchell]

I will be conducting a bacterial metagenomics survey with a recently acquired GS Junior sequencing platform. The Roche document titled "GS Junior System Guidelines for Amplicon Experimental Design" suggests that one way reads would be sufficient for metagenomics applications, and I would be inclined to agree, however I have yet to find a publication using the one-way Lib-L method. Theoretically one-way sequencing would produce reads up to 800bp.

Literature suggests a 400bp sequence (two-way read) could theoretically surpass 99.75% accuracy if low quality reads are excluded. Can anybody tell me the accuracy of a one-way 800bp read on the GS Junior?

[flxlex]

Theoretically one-way sequencing would produce reads up to 800bp

With the current chemistry, reads will not go very much beyond 500 bases. Where does your theoretical 800 bases come from?

[emitchell]

From "GS Junior Guidelines for Amplicon Experimental Design"

Basic considerations 1.1

• The optimal size range to obtain high quality (HQ) reads is 200-600 bp, including the Adaptor sequences. Concerning Amplicon length, note also that:
• Amplicons must be shorter than 800 bp or they will amplify poorly during emPCR amplification.
• Long range PCR products (see section 6) must be longer than 1500 bp or nebulization may be suboptimal, which could result in coverage bias.
• Therefore, Amplicons in the 800 – 1500 bp range are not recommended for these procedures.

Roche recommends one-way sequencing in metagenomics experiments where the likelihood of encountering the reverse read is low due to diversity in the sample. They also recommend designing primers that will push your amplicon length to as close to 800bp as possible without going over, suggesting that information will be acquired beyond the 500bp margin. The Lib-L em-PCR kit will only allow sequencing to proceed from the forward primer therefore I will not have a reverse read to increase accuracy at the tail end where basecall accuracy tends to drop off.

I may be misinterpreting the documentation. Even so my primers and products will be within the guidelines laid out by Roche yet there is no information on the accuracy of a one-way read using GS Junior titanium chemistry.

[kmcarr]

Unlike Sanger sequencing where the read length is determined randomly by incorporation of a di-deoxynucleotide, next generation sequencing methods operate for a fixed number of incorporation cycles which will limit the maximum length of a read you can obtain.

Current recipes for the GS-FLX and GS-Junior (I think) only run for 800 nucleotide flows, or 200 cycles of TACG. If the base composition of your sequence is well balanced you can expect to incorporate ~2.5 bases per cycle. This translates to final raw read lengths of ~500nt. If the base composition of your target is skewed, meaning in more homopolymer stretches, your raw reads will be longer. Only sequences with extremely long homopolymer runs will produce reads approaching 800nt.

[cystron]

amplicons over 600 bases (total length after adaptors added) rarely work well

The junior uses the same 200 cycles of the bases as the GS FLX so 800 flows.

I think the Metagenomics preference for uni-directional sequencing has more to do with downstream bio-informatics considerations than anything else.

[RCJK]

With the current sequencing kits, amplicons should definitely be <600 bp if possible. They've recently released a document describing how you need to modify the emPCR for amplicons larger than ~550-600bp to improve the number of reads that pass filter. According to that document, ~400bp would be the ideal amplicon length. That said, I have had some good results that were Lib-L amplicons of ~500-550bp (around 540 i think).
I have also been told that when they finally launch the longer read seq kits, it'll be for shotgun libraries only at first. At that point you could probably try to do longer amplicons, but I imagine the quality will suffer towards the end of the read and the software won't be optimized for it yet.