RNAseq - low cluster density - possible inhibitor?

[OnUrMark]

Greetings,
We have been getting low cluster densities, and as we increase the concentration, we are getting less clusters/tile/pM. It looks like there is no way for our samples to reach the optimal density of 400,000. It would seem that there is something wrong with our RNA-seq samples other than concentration that is causing this trend.
Ex:
9.6pM – 175,000 cl/tile (18,000 cl/tl/pM) – one lane
10pM – 146,000 cl/tile (15,000 cl/tl/pM) – one lane
20pM – 279,000 cl/tile (14,000 cl/tl/pM) – one lane
22pM – 261,000 cl/tile (12,000 cl/tl/pM) – one lane
25pM – 120,000 to 240,000 cl/tl (5,000 to 10,000 cl/tl/pM) – seven lanes, seven diff samples

We are multiplexing – 6 barcodes per lane. Samples were quantified with the Qubit flourometer. The bioanalyzer showed multiple peaks for our earlier samples that were daisy-chained. The extra peaks were eliminated when we changed the gel extraction protocol from 50C to RT, and the bioanalyzer showed very nice peaks at approx 350bp. Recently, one of our samples was checked with qPCR and found to be ¼ the concentration of the Qubit reading. It had previously been sequenced at 25pM and yielded 150,000 cl/tl.

1) Does anyone get such drastically different concentration readings?
2) Any ideas as to why so little of the sample is sequencing when the bioanalyzer shows our library is the target length in bp and PCR enrichment works well?
3) Any ideas as to why higher concentrations are not increasing (often decreasing) the cluster density?

I greatly appreciate any ideas or help you may have on these problems. Thanks, thanks and thanks!!

[csc]

In terms of increasing concentrations leading to lower densities: Are you correctly following Illumina's recommendations for high concentration library denaturation/dilution? Excess hydroxide carryover from the denaturation step certainly can lead to this phenomenon.

[OnUrMark]

We use the core facilities here for ngs. I have not looked at the high concentration directions from Illumina, but it is my understanding that the core facility dilutes samples so that the NaOH concentration is less than 0.8mN. Whether or not this is the "recommendation" I do not know.

[ECO]

What technology and kit are you using for your final library purification?

[OnUrMark]

We are using the Qiagen gel extraction kit with minielute columns. We have changed our gel several times, but presently we are using a 2% NuSieve 3:1 agarose gel (with EtBr) at RT for 80min.

[ECO]

Post final PCR as well? Or no final PCR?

[OnUrMark]

Yes. We PCR enrich the 350bp cDNA 12-18 cycles and then gel extract again. This final sample is checked on the bioanalyzer and the qubit before mixing with other samples for multiplexing and sequencing.

[John12824]

What technique are you using to add the barcodes?

We found that the 3 primer-based barcoding strategy from Illumina was very inefficient and many products lacked both flow cell attachment sites (P5/P7) and therefore could not undergo bridge amplification. If you are using this approach, if you run the PCR enriched product on the Bioanalyzer (before second gel extraction) you will likely see two peaks separated by 30 bp.

[OnUrMark]

Thanks very much for your suggestion. Although we don't use the 3 primer-based barcoding method, it may be informative to run pre gel extracted product on the bioanalyzer.

We ligate barcoded Y-adapters to cDNA. We run the "smear" on electrophoresis gel and extract at 350bp. Then, we PCR enrich with PE 1.0 and PE 2.0. I have always assumed that getting lots of product after PCR enrichment meant the adapters had annealed. Is there something else that could be happening?

[kshankar]

I realize this thread is a little old, but we are having trouble with getting our cluster densities up. A little background. We have made non-multiplexed RNA-seq libraries using PE primer 1.0 and 2.0 (synthesized from IDT) and get very high clusters/mm2.
However, with indexed libraries (modified Illumina's design using two primers, where the PCR 2.0 primer and the indexing primer are synthesized as one single primer) the clusters/mm2 are around (250 -300 K). We use qPCR to quantitate the libraries (after gel purification) and use 9 pM library DNA for cluster on the c-Bot.
Any hints on how to improve the cluster densities and explain the discrepancy between indexed and non-indexed libraries?

[OnUrMark]

Hello kshankar, I too would like to know why indexed libraries may produce lower cluster densities. How many samples per lane are you indexing?

Also, I still believe there are some contaminants from my gel extraction that are inhibiting the sequencing reaction. Can you tell me what gel percentage, brand, buffer and which kit you are using for your final gel extraction? Thanks so much.

[rocksd]

I realize this thread is a little old, but we are having trouble with getting our cluster densities up. A little background. We have made non-multiplexed RNA-seq libraries using PE primer 1.0 and 2.0 (synthesized from IDT) and get very high clusters/mm2.
However, with indexed libraries (modified Illumina's design using two primers, where the PCR 2.0 primer and the indexing primer are synthesized as one single primer) the clusters/mm2 are around (250 -300 K). We use qPCR to quantitate the libraries (after gel purification) and use 9 pM library DNA for cluster on the c-Bot.
Any hints on how to improve the cluster densities and explain the discrepancy between indexed and non-indexed libraries?

One possibility is the design of your index sequences. How long are your index sequences and how different they are? How many samples did you pool into one lane? What version of Illumina SCS software were you using? Can you let us know your index sequences?

I know that SCS1.8 identifies clusters based on the result of the first 5 or 6 cycles. Poorly designed indexes may lead to SCS to treat clusters close to each other as one cluster initially but eventually fail those clusters due to poor phasing/prephasing, hence leading to low cluster density.