We're multiplexing quite variable PCR products from environmental samples in single runs on a 454 Junior. Increasing the multiplex level to the most cost effective (we've got lots of samples!) means making sure you get a similar number of reads for each sample. As it's a Junior and we get only 70,000 reads a run when we're cooking, at the higher numbers of samples it can mean getting no reads (after QC and removal of bad reads) for particular samples in the run.
Initially I was carefully quantifying with PicoGreen and Invitrogen's Quant-it kits and diluting pools of products based on that. Typically amplicons range in abundance by 3 orders of magnitude, and (unintuitively) the low abundance amplicons tend to have many more reads. I wonder whether the kit is underestimating the amount of product (standard curves are linear with reproducibly good R^2 values)? The amplicons that are most abundant, and therefore diluted more are consistently the least represented in reads.
My next approach has been to quantify all amplicons, dilute to a similar order of magnitude, re-quantify and dilute them all to the correct molecules/µl input. Which still shows significant variability and is mind-numbingly tedious (and therefore error-prone!).
What other strategies do people use for normalization?
Does anyone have experience of the SequalPrep plates from Invitrogen? Do they work as they say? I would expect them not to be good with low abundance products (for some samples I pool several PCR reactions in order to get sufficient product to sequence, but I could pre-concentrate with an Ampure Xp purification, prior to using the SequalPrep plate?).
Thanks for your help
Initially I was carefully quantifying with PicoGreen and Invitrogen's Quant-it kits and diluting pools of products based on that. Typically amplicons range in abundance by 3 orders of magnitude, and (unintuitively) the low abundance amplicons tend to have many more reads. I wonder whether the kit is underestimating the amount of product (standard curves are linear with reproducibly good R^2 values)? The amplicons that are most abundant, and therefore diluted more are consistently the least represented in reads.
My next approach has been to quantify all amplicons, dilute to a similar order of magnitude, re-quantify and dilute them all to the correct molecules/µl input. Which still shows significant variability and is mind-numbingly tedious (and therefore error-prone!).
What other strategies do people use for normalization?
Does anyone have experience of the SequalPrep plates from Invitrogen? Do they work as they say? I would expect them not to be good with low abundance products (for some samples I pool several PCR reactions in order to get sufficient product to sequence, but I could pre-concentrate with an Ampure Xp purification, prior to using the SequalPrep plate?).
Thanks for your help