Sample Eveness

[george_orwell]

Hi all

I currently use the MiSeq for 16s metagenomics using the V3-V4 primers from Caporaso et al (2010). I've finally nailed down library quantification and optimal cluster generation, but I've been having issues with sample evenness.

Currently I'm using PicoGreen to quantify my samples individually and then pooling equimolarly for qPCR. The range for sampling depth on my samples (~150 per run) ranges from 80k-350k. While I'm happy to be getting great quality data, I'd like to nail down the eveness on the MiSeq.

Anyone have any suggestion for optimal evenness?

Thanks,

Pat

[kmcarr]

Hi all

I currently use the MiSeq for 16s metagenomics using the V3-V4 primers from Caporaso et al (2010). I've finally nailed down library quantification and optimal cluster generation, but I've been having issues with sample evenness.

Currently I'm using PicoGreen to quantify my samples individually and then pooling equimolarly for qPCR. The range for sampling depth on my samples (~150 per run) ranges from 80k-350k. While I'm happy to be getting great quality data, I'd like to nail down the eveness on the MiSeq.

Anyone have any suggestion for optimal evenness?

Thanks,

Pat

Pat,

My suggestion is don't drive yourself crazy. Are 80K reads enough to do descent metagenomics? Almost certainly yes. You are never going to achieve the type of library balance for 150 16S samples like you can with, for example, 12 TruSeq libraries.

Our lab does a large number of 16S projects (mostly V4 only [515f/806r] but also some V3-V4 [357f/806r]) and initially they were doing picogreen on every reaction and trying to pool an equimolar amount. Sometimes it worked well, other times less well. This was also laborious as hell. The lab switched to using Life Technologies SequalPrep Normalization Plate Kit (I believe this was originally an Invitrogen product before Life Tech swallowed them whole). The idea is that you put an excess of PCR product in each well of the plate, the wells bind a fixed amount, after removing the unbound portion and washing you elute, pool and concentrate what was bound. Do you get a perfectly even number of reads from each library? No. Do you get reasonable distribution of reads, with a sufficient number from each to meet your experimental objective? Yes.