RNA-SEQ (human transcriptome) throughput

[vinay052003]

Hi everyone,
We are looking at different sequencing technologies to sequence human transcriptome. We are mainly interested in Illumina HiSeq. Does anyone have idea about the throughput for human transcriptome. We are planning to perform 100-bp paired end (100-100). The ideal throughput from HiSeq is there on the Illumina website... I just want get the idea about practical/actual throughput from the sequencer.

[GenoMax]

If I am understanding your question correctly:

It is going to depend on how good your libraries are, how well your cluster station(s)/sequencer(s) run. It is not uncommon to get between upwards of 200+ million reads per lane from a HiSeq flowcell that is using v.3 chemistry.

[james hadfield]

A PE100bp run with a multipex read takes about 11 days for a pair of flowcells. Assuming 200M reads per lane and 50M reads needed per sample you could run 64 samples every 11 days.

[kmcarr]

Hi everyone,
We are looking at different sequencing technologies to sequence human transcriptome. We are mainly interested in Illumina HiSeq. Does anyone have idea about the throughput for human transcriptome. We are planning to perform 100-bp paired end (100-100). The ideal throughput from HiSeq is there on the Illumina website... I just want get the idea about practical/actual throughput from the sequencer.

Do you really need to do 100bp PE reads for the human transcriptome? If the application is an expression study (transcript counting) you only need enough sequence to confidently assign a read to the correct gene. Yes, for highly similar paralogs this discrimination may require 100bp PE but these are the edge cases. For expression studies in well annotated model organisms I typically will recommend 75bp SE reads to researchers. I feel the marginal increase in mapability beyond that is not worth the additional cost.

[epistatic]

I wouldn't recommend PE-100 for TruSeq RNA libraries unless you are altering the insert size. You will have overlapping reads (negative insert). There is a thread here about negative insert sizes, combining overlapping reads into a long single read for mapping, etc. We find SR-50 is great for most applications, PE-50 is good for chimeric transcripts, maybe PE-75 if you are willing to pay for it.