Announcement

Collapse
No announcement yet.

'Assembly PCR' for homemade amplicon library prep

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • 'Assembly PCR' for homemade amplicon library prep

    I'm trying to make homemade amplicon library preps for the Illumina MiSeq, and my fusion primers always end up being around 100bp (if not more!) by the time I add all the necessary sequences (p5/p7, indices, primer landing sites, N's for diversity, etc.). This requires I buy oligos at the 250nm scale of synthesis rather than the standard 25nm scale. When I'm buying a whole plate of 96 indices, this gets very expensive very fast!

    I'm thinking of an 'assembly PCR' approach in which I order four 25nm oligos. Two short ones that contain the sequence-specific sequence fused to the Illumina sequencing primer landing sites and two additional short ones that contain the p5/p7 sequences, indices, and a bit of the primer landing site. These could all be thrown into a single PCR reaction (perhaps with a bias towards the 'outer' primers containing the p5/p7 sequence).

    Of course I would have to pay attention to strandedness and directionality when designing the oligos, but I wanted to see if anyone has tried this before I spend too much time thinking about it.

    This idea would also have the benefit of being able to change out the sequence-specific 'inner' primer and re-use the outer primer/indices for different experiments.

  • #2
    I think Illumina recommends a two-step approach. Fusion primers with barcode and sequencing primer locations, then a second PCR with P5/P7 ends being added. If you do it all at once it may get ugly. Also if you are low diversity adding in a N+1, N+2 linker in there greatly helps diversity.

    Comment


    • #3
      Try this:

      http://supportres.illumina.com/docum...15044223-b.pdf

      Also, I'm not sure you need the extra N bases to increase diversity, if you're using the newest MiSeq software.

      Cheers,

      Scott.

      Comment


      • #4
        Hi esherman,

        A few options/opinions to offer on this.
        1) If you do decide to go 2-step you risk contamination. These workflows are really really hard to keep clean. Personally I 'hate' this approach but it does kind of depend on what you are using the data for.
        2) Do you need to do paired end sequencing? - we are getting good single direction sequencing out to 250+ bases (on v2 300 nano kits), so if your amplicon is shorter than that you could consider it (i.e you only need a sequencing primer at one end).
        3) Use a custom sequencing primer (e.g one with LNA's) to shorten the the total length of the primer you need. Search some of the seq-answer threads on this for a few hints or PM me.
        4) a combination of points 2+3 above is what we (mostly) use to keep primer sizes down which helps cost and PCR efficiency.

        best of luck with your workflows, Cheers Mike
        Last edited by bunce; 11-13-2014, 06:16 PM.

        Comment


        • #5
          I'd second the two-step approach. Do a first round with a fusion primer with the specific sequence and full/partial read primer sequence followed by a second round that adds the indexes and p5/p7.

          That way gives you the most flexibility, as well, because you can keep reusing the indexing primers for future experiments and can always switch around which library gets which index so you don't end up with incompatible overlap.

          As for cross contamination, having separate pre and post PCR benches is critical (much more common in corporate labs), but you shouldn't need to do many PCR cycles at all to add the p5/p7 and indexes, so you really just need to make sure your first PCR product and final libraries stay the hell away from the bench where you set up the first round of PCR.

          Comment


          • #6
            Originally posted by cmbetts View Post
            As for cross contamination, having separate pre and post PCR benches is critical (much more common in corporate labs), but you shouldn't need to do many PCR cycles at all to add the p5/p7 and indexes, so you really just need to make sure your first PCR product and final libraries stay the hell away from the bench where you set up the first round of PCR.
            I would not be so hasty in talking down the contamination risk in amplicon workflows - our ancient DNA set-up is 'extreme' and we still get bleed-though of tags (we never re-use index pairs). If you are doing a 2nd round PCR you still need to open that tube in a post-PCR 'area' to spike in the 1st round PCR. Amplifying amplified DNA that then undergoes another amplification (cluster generation) can cause headaches downstream from chimera formation to contamination that builds over time. But, again, it does depend on the end use of the data. thats my 2cents worth - cheers, Mike

            Comment


            • #7
              As a rule of thumb, we never use the same barcode combination twice on a plate. It's not like you have the option of running the same barcodes in multiple lanes like you can on a HiSeq.

              For us the most critical part is keeping your primer stocks contamination free prior to the PCR. After normalization and pooling you would have no way of knowing your DNA was tagged with multiple barcodes without some sort of internal control or verification of the target.

              Pipette wisely!
              Last edited by DNA_Dan; 11-13-2014, 08:49 PM.

              Comment


              • #8
                Originally posted by bunce View Post
                I would not be so hasty in talking down the contamination risk in amplicon workflows - our ancient DNA set-up is 'extreme' and we still get bleed-though of tags (we never re-use index pairs).
                I wonder if you have considered possibility of carry over from a previous run on MiSeq. We use two step PCR and it works well. In addition, it is convenient and we can pool and sequence other libraries as well without risk of unwanted interaction between custom and Illumina sequencing primers. Nano is the most expensive option for Illumina systems per Gb of data.

                Comment


                • #9
                  Originally posted by nucacidhunter View Post
                  I wonder if you have considered possibility of carry over from a previous run on MiSeq. We use two step PCR and it works well. In addition, it is convenient and we can pool and sequence other libraries as well without risk of unwanted interaction between custom and Illumina sequencing primers. Nano is the most expensive option for Illumina systems per Gb of data.
                  Hi, Sure - carry-over is always a possibility (we have been using the bleach wash protocol since february). I suspect (with no evidence) that people may 'blame' carry-over when it is on fact the lab workflow causing issues.

                  I don't doubt that a 2-step PCR protocol works well - it will always generate a truck-load of library. The question is what portion or reads are chimeric or have tags that you have not used on that run? I would be interested if anyone (using 2-step PCR and re-using indexes), has quantified this?

                  Comment


                  • #10
                    Originally posted by bunce View Post
                    Hi, Sure - carry-over is always a possibility (we have been using the bleach wash protocol since february). I suspect (with no evidence) that people may 'blame' carry-over when it is on fact the lab workflow causing issues.
                    Illumina has acknowledged carry over issues: https://icom.illumina.com/MyIllumina...applications-m

                    If by “evidence” you mean how many people look for evidence of contamination source in their reads, I do not have an answer.

                    I don't doubt that a 2-step PCR protocol works well - it will always generate a truck-load of library. The question is what portion or reads are chimeric or have tags that you have not used on that run? I would be interested if anyone (using 2-step PCR and re-using indexes), has quantified this?
                    Two step PCR if done properly does not generate truck load of amplicons. The aim should be number of cycles to generate enough yields for a QC and 3-4 sequencing runs which usually would be 50 ng>.

                    Chimeric reads actually would be more prevalent in one step PCR rather than two step, because two major reasons for Chimera formation during PCR are incomplete extension and strand invasion. The condition for both are favoured in one step PCR where the reagents are most likely to deplete and concentration of amplicons increase to a critical point favouring those reactions.

                    By using indexed primers for second step from a plate sealed with a plastic film and frozen, thawed and resealed multiple times, I have got 1 in 10,000 reads that had indices not used in the reactions. By using individual tubes containing an indexed primer and opening them one at a time such as recommended in Nextera protocol, no read was detected that had unused indexed primer. One obvious advantage using two step PCR with dual indexing (such as one described in Illumina 16S sequencing protocol, see link in #3) is that identifying and eliminating cross contamination post-sequencing (caused either by physical contamination during library prep or image analysis error with higher cluster densities) would be easy as chance of both primers being contaminated is reduced.

                    Amplifying amplified DNA that then undergoes another amplification (cluster generation) can cause headaches downstream from chimera formation to contamination that builds over time. But, again, it does depend on the end use of the data.
                    Cluster generation should not encourage chimera formation because bridge amplification takes place in solid state and unlike in solution PCR, fragments are not free to interact with each other. Fragments that are in close proximity or hybridize to each other and may fuse and form a cluster, will not produce pure signal either because of presence of mix template amplification or multiple primer binding sites and will fail filter so they cannot contribute to chimera reads.

                    It is not clear to me what sort of contamination would built over time, if one follows standard molecular biology lab hygiene such as using disposable gloves, aerosol free tips and cleaning bench and pipette surfaces.

                    Comment


                    • #11
                      Hi nucacidhunter, I guess we will agree to disagree on some of these points. I can't conceive of a way in which a 2-step protocol can be 'cleaner' or less susceptible to artefacts than a 1-step set-up. It may be cheaper, arguably more efficient? - but not cleaner. When it comes to NGS, PCR is a necessary evil it should be minimised in anyway possible.

                      Originally posted by nucacidhunter View Post
                      Chimeric reads actually would be more prevalent in one step PCR rather than two step, because two major reasons for Chimera formation during PCR are incomplete extension and strand invasion. The condition for both are favoured in one step PCR where the reagents are most likely to deplete and concentration of amplicons increase to a critical point favouring those reactions.
                      Your logic is lost on me here as the 2-step method also goes though a 1st round PCR that is just is susceptible to Chimeras (I think we agree that PCR cycles should be kept to a minimum). In amplicon sequencing some people work indexes into their 1st round PCRs then pool and amplify (as a pool) to get p5-rd1/rd2-p7 onto the products. In these situations there has been a number of reports of 'jumping' indexes presumably the result of incomplete extension. This "can't" happen (i.e. index jumping) in a 1-step workflow as the forward and reverse indexes are the only ones in tube.

                      Originally posted by nucacidhunter View Post
                      By using indexed primers for second step from a plate sealed with a plastic film and frozen, thawed and resealed multiple times, I have got 1 in 10,000 reads that had indices not used in the reactions. By using individual tubes containing an indexed primer and opening them one at a time such as recommended in Nextera protocol, no read was detected that had unused indexed primer.
                      If a 1/10,000 contamination rate suits your application then that is good - it won't suit everyone. People not as adept at removing those pesky films may report a higher rate????

                      Originally posted by nucacidhunter View Post
                      One obvious advantage using two step PCR with dual indexing (such as one described in Illumina 16S sequencing protocol, see link in #3) is that identifying and eliminating cross contamination post-sequencing (caused either by physical contamination during library prep or image analysis error with higher cluster densities) would be easy as chance of both primers being contaminated is reduced.
                      This is is a way to identify contamination but is not an "obvious advantage " over a 1-step library generation. A 1-step workflow that integrates indexes and adapters at the 1st PCR so contamination is just as easy to spot.


                      Originally posted by nucacidhunter View Post
                      Cluster generation should not encourage chimera formation because bridge amplification takes place in solid state and unlike in solution PCR, fragments are not free to interact with each other. Fragments that are in close proximity or hybridize to each other and may fuse and form a cluster, will not produce pure signal either because of presence of mix template amplification or multiple primer binding sites and will fail filter so they cannot contribute to chimera reads.
                      Agreed, chimera's are not really an issue at cluster stage. But it is an amplification stage so one contaminatiing template molecule could initiate that cluster.


                      Originally posted by nucacidhunter View Post
                      It is not clear to me what sort of contamination would built over time, if one follows standard molecular biology lab hygiene such as using disposable gloves, aerosol free tips and cleaning bench and pipette surfaces.
                      .

                      Aerosols build in labs over time. In a post PCR area you can use ART tips, UV and gloves but this minimises contamination it won't remove it. A good PCR once 'opened' will generate aerosols with 10^5-10^9 copies of DNA that can travel through HEPA filters and build in a lab.

                      But to bring this banter back to esherman's question - there are good ways to generate amplicon libraries using both 1-step and 2-step methods there are strengths and drawbacks to both approaches. How you tackle this is very much dependent on budget, sensitivity, contamination concerns and the application you intend for the data.

                      Comment


                      • #12
                        Hi bunce, I am fine with disagreement as far as it is evidence based and I agree with you in that cleanliness or tolerance of contamination is dependent on application and other factors. In two-step PCR usually first PCR goes for 15-20 cycles and then an aliquot is used for second PCR, say for another 10 cycles. This would generate fewer artefacts than a similar one-step PCR cycled 30 times because the amplicons would have less concentration and PCR reagents are replenished reducing the chances of incomplete extension and high concentration of amplicons two major cause of PCR artefacts.

                        The level of cleanliness you are hinting is more relevant to forensic and ancient DNA work which unfortunately most of the time suffers from contamination during or even before sample collection.

                        Out of interest is there any evidence that how many of those 10^5-10^9 aerosol copies of DNA that travel through HEPA filters and build in a lab, contaminate the work being done in that lab.

                        Comment


                        • #13
                          Seems to me that one of the biggest 'deterrents' of contamination between samples within one prep, between preps across time, etc is to use enough template DNA so that any contamination won't be preferentially amplified.

                          Our platform may be more-ammenable to this type of prep, as we know our distal ends of the DNA to be sequenced…

                          We use a 2-stage PCR to build out our Illumina ends. As suggested earlier, the primary PCR uses library-specific oligos, but with about 1/2 of the Illumina adapters extending off the ends.
                          The Secondary PCR primers introduce the P5/P7 ends, as well as the barcode, but have no complementarity to the initial library, only the distal ends of the primary PCR product that are part of the illumina adapter just inbound of the barcodes.

                          We perform a PCR clean-up between each step to clean out the primers…
                          We use 10ng of input DNA into a 50ul PCR reaction, and that amount of template is a ton more than any contamination event unless you're an idiot...
                          Also, in-line PCR negative controls are always run…
                          And we don't ever sequence with the same barcodes two runs in a row...
                          And I am planning on installing the Bleach wash, but we typically only see about 1% of the total yield of DNA going into the 'undetermined' fastQ file.

                          I got fancy once, and added 1/1000th the amount of primary PCR primers for 2 cycles, and then added in the typical amount of secondary primer (barcode, P5P7)…that works but I don't like opening fresh PCR products and adding more primers to do more PCR. now that's asking for contamination.

                          Comment


                          • #14
                            Originally posted by ZAAB View Post
                            Seems to me that one of the biggest 'deterrents' of contamination between samples within one prep, between preps across time, etc [U]is to use enough template DNA so that any contamination won't be preferentially amplified
                            .

                            I agree - it is a numbers game. The issue however is that some applications are setup to find rare variants (e.g. circulating cancer cells or low concentrations of microbes in a complex mixture) - in these scenarios a low level (and unpredictable) baseline of contamination can be problematic. Again, (to sound like a broken record) the workflow - be it 1-step and 2-step - depends on end-use of the data.

                            The point I am trying to make here is that contamination in NGS workflows is real and needs to be managed - never more so than when amplifying amplified DNA.

                            Comment


                            • #15
                              Contamination aside, have any of you looked at the effect of resampling bias from doing so much PCR on a sample? I have a customer looking to do this approach coming from RNA. This means 4 rounds of PCR, 1 RT reaction, 1 target specific nest, then the Illumina 2 step. I plan on titrating yield on a Lonza gel to optimize PCR cycles, but still sounds like a lot of resampling.

                              Comment

                              Working...
                              X