Seqanswers Leaderboard Ad

Collapse

Announcement

Collapse
No announcement yet.
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Adventures in cluster density titration.

    We have perennial issues with obtaining sane/consistent library concentrations prior to clustering. Those of you with a HiSeq will probably have this same problem, intermittently, if not persistently.

    We had a particularly vexing set recently.

    These all were TruSeq RNA prep libraries with a few protocol changes. Briefly, since we often do 2x101 PE de novo RNA seq, we usually back off on the fragmentation time to generate longer inserts. Then, to compensate we also use 0.8x Ampures in most places in the protocol. Just this batch we added a single, "hi-cut" Ampure, where we precipitated that part of the library that came down with a 0.5x Ampure cut and then recovered the molecules left in the supernatant (with more Ampure).

    Also, we do 1/2 reactions. Pretty much just cut the volume of everything in half. (We started doing this because the multi-month back-orders for these kits made them too precious to waste.) Also, 15 cycles of enrichment PCR is way too high. We always cut that back. In this case to 10 cycles.

    Anyway, the TruSeq RNA kit is a champ and almost always delivers a large library. We did per-library qPCRs and made pools based on the concentrations we obtained. I'll spare you the gory details. (And they are gory.) Each pool has (ostensibly) equal contribution of anywhere from 8 to 14 libraries. So this post goes into a little detail about 4 pools, their titrations and final cluster densities.

    Here are the Agilent DNA High Sensitivity chip results for the 4 pools:









    Here is a table of results:



    "len" is just the length as called by the Agilent Chip. (But you can see from the images that most of the library is in the 250-550 bp range.) Since all the assays we use are "mass" assays we do a correction for size, based on this number.

    The next 3 columns are 3 different methods of estimating the concentration of the libraries. The first is just what the Agilent chip estimates. "qPCR nM" is what we calculate using a KAPA kit, but with an Illumina phiX library as the standard. Our high standard is 1 nM just to give you an idea of where we are in the concentration scale. For fluorimetry we take an aliquot of the sample and denature it (95 oC for 3 minutes, followed by chilling on ice) and then use the ribo-green fluor with its included rRNA standard. This is weird, I know, but the idea is that lots of your library might be single stranded, so you want that counted. Double-stranded (pico-green) fluorimetry will not show the single stranded stuff. Alas, pico-green's reading for double-stranded molecules is not the same as for single-stranded molecules. But I think this assay will be less confounded by mixtures of single-stranded and double-stranded molecules.


    Distressing that both qPCR and fluorimetry give higher estimates of the concentration. Anyway, we decided to go with the fluorimetry values for making what we called "presumptive 2 nM" stocks. Our experience is that the larger the dilutions necessary to read 2 nM, the further from 2 nM they end up being upon re-quantification.

    Then we take the "2 nM" stocks and do a final qPCR on them. They are all high. Then we dilute them again to what will hopefully be 2 nM based on those values. (We use 0.1% Tween as a surfactant in our dilutant.) Then fire up a run, loading a single pool per lane in the flow cell at 14 pM. 750-850 is spec for v3 chemistry. But our HiSeq seems to give excellent results even when the cluster density is above 1000. So we would rather be higher -- especially for the last 2 pools. Still, nothing to complain about. Except for the enormous time and effort it took to dial the concentrations in.

    As I mention above, I left out details about the initial qPCR to create the pools. Also, the early qPCR results were so wacky I decided to "buy a vowel" by doing a 50 cycle MiSeq run on 2 of the pools (pools 2 and 3). Using the result of those we ended up at 907 and 1050 K/mm^2 cluster densities -- exactly where we wanted to be. I only did 2 of the pools on the MiSeq because any other combination of pools resulted in index overlaps.

    Since we routinely have >50 libraries per flow cell, doing a MiSeq run for titration is probably not in the cards until we switch to dual indexing -- which would give us at least 96 unique index pairs.

    Any comments?
    --
    Phillip

  • #2
    I heard that illumina is coming out with a QC kit for titrating libraries on the MiSeq. I guess they are listening to all the complaints about clustering concentrations and wasting runs. Don't know any details on the kit though. Your qPCR v bioanalyzer discrepancies look a lot like ours.

    Comment


    • #3
      We can deal with qPCR vs bioanalyzer discrepancies, what is killing us are different qPCR results with different starting concentrations of library. Or just, crazy, variable results qPCR to qPCR, even with the same concentration library. The latter has been less of an issue since we started doing a hard-spin on all libraries prior to do a qPCR -- the idea being that there may be microscopic debris in the libraries (dust, beads, etc.) that cause the mixture to be non-homogeneous.

      However the former is something we need to get our heads around. We are using Illumina phiX libraries as our standard. The log-linear distributions on these give very high R^2 values when we check them (> .95 typically). Nevertheless, it is clear that the phiX intensity/dilution slope is not the same as our sample slope. Generally replicates at the same dilution give the same qPCR result. But at different dilutions we get a different results. Usually.

      I may just be missing something very obvious to someone who has formal qPCR training.

      --
      Phillip

      Comment


      • #4
        Could this be within the realm of pipette variability? Are you working with small-ish volumes?

        Comment


        • #5
          I have done a lot of TruSeq RNA and I find that I can get very standard clustering using mass measurement on the QuBit. This has been my go to for quite a while, I still do qPCR but this mass off the QuBit is a measurement that has been consistent for me. In fact this is true for all of my RNA-seq libraries, older whole transcriptome and mRNA-seq, all the Tru-seq, 3' directional, and dUTP.

          I just use the assumption that 2ng/ul equals 10 nM and I get consistent clustering everytime on our GA and have even used this for libraries sent to MiSeq and HiSeq.

          Comment


          • #6
            Originally posted by mnkyboy View Post
            I have done a lot of TruSeq RNA and I find that I can get very standard clustering using mass measurement on the QuBit. This has been my go to for quite a while, I still do qPCR but this mass off the QuBit is a measurement that has been consistent for me. In fact this is true for all of my RNA-seq libraries, older whole transcriptome and mRNA-seq, all the Tru-seq, 3' directional, and dUTP.

            I just use the assumption that 2ng/ul equals 10 nM and I get consistent clustering everytime on our GA and have even used this for libraries sent to MiSeq and HiSeq.
            Any adjustments needed for DNA libraries? Are your library insert size distributions narrow? (Less than 100 bp)?

            --
            Phillip

            Comment


            • #7
              Originally posted by GW_OK View Post
              Could this be within the realm of pipette variability? Are you working with small-ish volumes?
              Yes, that could be a possibility. Except the same pipette is probably being used for the phiX dilutions as well as the sample library dilutions.

              I was thinking there might be a difference in the amplification efficiency for different library types. The phiX library has an average size of 500 bp. These RNA seq libraries are 200-700 bp. Their GC composition may be different as well

              --
              Phillip

              Comment


              • #8
                Originally posted by pmiguel View Post
                Any adjustments needed for DNA libraries? Are your library insert size distributions narrow? (Less than 100 bp)?

                --
                Phillip
                We do a tiny fraction of DNA compared to RNA-seq. The mass conversion has worked for these, but I do take into account the size if they are different from our standard RNA libraries and rely on the qPCR more as a double check for the mass in relation to size and to check ligation of adapters. Most of the DNA libraries we do not make and we have had issues with users not getting efficient ligation but still have measurable DNA. This has not been an issue with the RNA libraries.

                Our DNA library sample size is a little small for me to have as much confidence as I do using the mass for the RNA-seq libraries.

                Comment


                • #9
                  We observe non-linear properties when the size distributions are large, as yours are. E.g., doubling the concentration for clustering from 4 pM to 8 pM yields less than 2X clusters, while halving the concentration produces more than 1/2X clusters. The phenomenon largely disappears with a tight size distribution. (Note that RNA-Seq libraries typically have a tight size range and, per mnkyboy, are usually well-behaved in this regard.) The reason for this phenomenon is unclear but, as a practical solution, we usually repeat the size selection.

                  Broad size distributions would be expected to produce even more complex behaviors in qPCR, as the amplification efficiences of the smallest vs. largest fragments are likely to vary significantly. For example, if the efficiency were 1.0 for a 300bp fragment and 0.9 for 600bp, then the smaller fragment would be ~five-fold enriched relative to the larger one after only 15 cycles and continue to increase thereafter. This would manifest itself as a decreasing delta Ct per cycle, such that more dilute samples (which require more cycles to reach the threshold) would yield a higher apparent concentration.

                  Comment


                  • #10
                    Thanks mnkyboy and HESmith, I think your comments address our issues.

                    Since we are situated in an Ag college of Purdue, which itself has no direct association with a human hospital, we do comparatively little standard RNAseq. Instead most of our projects are de novo transcriptome. Hence our RNA libraries tend to look more like non-upper size cut selected DNA libraries.

                    We just started trying doing Ampure upper cuts. These appear to work, just need a little fine tuning.

                    --
                    Phillip

                    Comment


                    • #11
                      high cut method

                      Any updates on this method?
                      I have tried it myself, results as shown in attached image. I like the 0.7 ratio best, maybe 0.65. Haven't used it for real libraries much yet, though. My protocol was to bind DNA to beads 15 minutes, place on magnet, remove 90% sup and purify with 1.25:1 fresh beads.
                      Attached Files

                      Comment

                      Latest Articles

                      Collapse

                      • seqadmin
                        Strategies for Sequencing Challenging Samples
                        by seqadmin


                        Despite advancements in sequencing platforms and related sample preparation technologies, certain sample types continue to present significant challenges that can compromise sequencing results. Pedro Echave, Senior Manager of the Global Business Segment at Revvity, explained that the success of a sequencing experiment ultimately depends on the amount and integrity of the nucleic acid template (RNA or DNA) obtained from a sample. “The better the quality of the nucleic acid isolated...
                        03-22-2024, 06:39 AM
                      • seqadmin
                        Techniques and Challenges in Conservation Genomics
                        by seqadmin



                        The field of conservation genomics centers on applying genomics technologies in support of conservation efforts and the preservation of biodiversity. This article features interviews with two researchers who showcase their innovative work and highlight the current state and future of conservation genomics.

                        Avian Conservation
                        Matthew DeSaix, a recent doctoral graduate from Kristen Ruegg’s lab at The University of Colorado, shared that most of his research...
                        03-08-2024, 10:41 AM

                      ad_right_rmr

                      Collapse

                      News

                      Collapse

                      Topics Statistics Last Post
                      Started by seqadmin, Yesterday, 06:37 PM
                      0 responses
                      7 views
                      0 likes
                      Last Post seqadmin  
                      Started by seqadmin, Yesterday, 06:07 PM
                      0 responses
                      7 views
                      0 likes
                      Last Post seqadmin  
                      Started by seqadmin, 03-22-2024, 10:03 AM
                      0 responses
                      49 views
                      0 likes
                      Last Post seqadmin  
                      Started by seqadmin, 03-21-2024, 07:32 AM
                      0 responses
                      66 views
                      0 likes
                      Last Post seqadmin  
                      Working...
                      X