HiSeq 2000 RNA-Seq Mapping Rates

[Lee Sam]

We're running and mapping human samples on our new(ish) HiSeq2000 and we're seeing some low mapping rates out of ELAND. Even if you don't run ELAND, I'd be interested in hearing what other HiSeq users' mapping rates look like with Bowtie/BWA/whatever.

[googleboyjay]

I have used bowtie to map illumina RNA-seqs to the original genome. The mapping rate is 6.98% without considering repeat mapping. The true rate is even lower.

[Lee Sam]

6% is kind of surprisingly low. We're seeing 20% on our transcriptome runs with Bowtie, but I suspect that may have something to do with the library construction and how the paired reads may overlap (which Bowtie can not deal with, minimum insert size is 0).

[googleboyjay]

6% is kind of surprisingly low. We're seeing 20% on our transcriptome runs with Bowtie, but I suspect that may have something to do with the library construction and how the paired reads may overlap (which Bowtie can not deal with, minimum insert size is 0).

I am currently not using PE. All available is a single fasta file of reads. I have tested two totally different bacteria species. Both of them are of this level, i.e, mapping rate is pretty low. I do not know the reason. Any suggestions?

[dsidote]

I realize the OP asked about the HiSeq. We don't have a HiSeq, but I wouldn't expect the results to be very different from the GAIIx since they use the same chemistry so I'll throw this out there for comparison.

We recently sequenced a maize transcriptome with 101bp single end reads. We truncated to 76 bp and used tophat, which uses bowtie, and we were able to map 66% of the reads.

I think something is wrong if you are only mapping 6%. Was the library ribo-depleted or polyA? Was the library checked on a bioanalyzer to see if primer-dimers were present and to assess how much rRNA was present in the sample? We have had user prepared libraries that have had as much as 80% rRNA present in the final library.

[kopi-o]

We have good mapping rates with HiSeq RNA-seq on human samples, >70% and up to ~80% with bwa or TopHat. (PE 2x100)

[NextGenSeq]

Is this mapping to the genome or the transcriptome (coding exons)?

Also, if your RNA has mycoplasma contamination (from cell culture) you will get much lower mapping rates.

[googleboyjay]

We have good mapping rates with HiSeq RNA-seq on human samples, >70% and up to ~80% with bwa or TopHat. (PE 2x100)

I am sorry. I made a mistake calculating the statistics. The mapping rate is 77.45%. However, most of the mapped sequences are duplicately mapped. I am not doing the sequencing part of research. I am doing the annotation part. So I am not clear of the details of the sequencing.

best wishes!

[googleboyjay]

Is this mapping to the genome or the transcriptome (coding exons)?

Also, if your RNA has mycoplasma contamination (from cell culture) you will get much lower mapping rates.

I am mapping the sequences to the original genome. I am not clear of the environment. I am doing downstream annoation.

[yh_gu]

Hi,
When using Bowtie, how to decide which seed length is best?