Bowtie tags for dUTP strand specifc RNA-seq

[BacanBan]

Hello,

I'm new to this area and I just wanted to confirm I'm not missing out on something important. I have RNA-seq strand specific, single end data generated using the duTP method. I want to align my raw reads to the reference genome and am using Bowtie2, version 2.0.2. I read the manual and these forums, but I'm still confused about which tag, if any, I have to supply to bowtie to let it know that I'm using strand specific data.

Reading the manual, makes me think --nofw/--norc, is not what I'm looking for, but I'm still unsure. I appreciate any help you can provide me with. And I'm sorry if the answer is really obvious.

Thanks.

[Kate.W]

Hi! Did you find the answer? I am stuck with that same uncertainty... Thanks!

[kmcarr]

Hello,

I'm new to this area and I just wanted to confirm I'm not missing out on something important. I have RNA-seq strand specific, single end data generated using the duTP method. I want to align my raw reads to the reference genome and am using Bowtie2, version 2.0.2. I read the manual and these forums, but I'm still confused about which tag, if any, I have to supply to bowtie to let it know that I'm using strand specific data.

Reading the manual, makes me think --nofw/--norc, is not what I'm looking for, but I'm still unsure. I appreciate any help you can provide me with. And I'm sorry if the answer is really obvious.

Thanks.

Bowtie by itself is not designed for aligning RNA-Seq data so it doesn't have any options to specify stranded RNA libraries. To align RNA-Seq data you should use TopHat which uses Bowtie but adds functionality specific to RNA-Seq. Using TopHat you would use the parameter "--library-type fr-firststrand" if your reads are from a strand specific library prepared using the dUTP method.

[BacanBan]

Hi,

I did try the bowtie way without any of the options just to see what it was like. I also tried the tophat with the option --library-type fr-firststrand, which is explained by kmcarr. I decided to go with the tophat one as it seems like it's better suited to the job. Though to be fair, their outputs were pretty similar.

[summere]

Bowtie Options for Strandedness

Hi-

Bowtie does have options for directionality. From the command line, just type "Bowtie" (without the quotes) and look at the entry for --nofw/--norc do not align to forward/reverse-complement reference strand.

Hope that helps!
Summer

[dpryan]

Hi-

Bowtie does have options for directionality. From the command line, just type "Bowtie" (without the quotes) and look at the entry for --nofw/--norc do not align to forward/reverse-complement reference strand.

Hope that helps!
Summer

Yeah, that wouldn't make much sense for RNAseq, though one shouldn't then be directly using bowtie anyway, at least unless your organism lacks splicing.

[summere]

Makes Sense for RNA-seq

I have de novo assemblies that I built using the Trinity pipeline, which was created by the Broad Institute and which uses an RSEM/Bowtie platform.

I used a modified pipeline based on RNA-seq protocols outlined in Nature; here's more information in case you want to read up:

http://www.nature.com/nprot/journal/v8/n8/full/nprot.2013.084.html

http://www.nature.com/nbt/journal/v29/n7/full/nbt.1883.html

[dpryan]

Assembling transcripts isn't exactly the standard usage for RNAseq. Sure, if that's what you're doing then it would be useful.

[tokikake]

Hi,
I have the same problem with bowtie2:

I have bacterial paired-end Illumina reads from a dUTP library. bowtie2 has flags for strand specificity: "summere" only mentioned the --nofw/--norc flag, but isn't the --fr/--rf flag the more important one??

In the past I only had single-end RNASeq reads and used to reverse complement the R1 and map it in SE mode with bowtie2. For the paired-end data I also rev/comp the R1 and combine both and map the in the SE mode. In both cases I do see stand specificity, but I think that this is not the proper way to do. bowtie2 handle them as unstranded reads and so they can be mapped on both strands and maybe cause higher background, or am I wrong?

So, is there anyone out there mapping bacterial dUTP RNASeq data with bowtie2? And what kind of flags do you set? Or what kind of program did you use despite bowtie2?
I tried all possible combinations with the --fr/--rf and --norc/--nofw flags and always one of the reads doesn't map at all and the other does not map in sense orientation.

Thanks a lot for advice or just discussion, so I don't have the feeling to spin round!

Sonja

[kmcarr]

Hi,
I have the same problem with bowtie2:

I have bacterial paired-end Illumina reads from a dUTP library. bowtie2 has flags for strand specificity: "summere" only mentioned the --nofw/--norc flag, but isn't the --fr/--rf flag the more important one??

In the past I only had single-end RNASeq reads and used to reverse complement the R1 and map it in SE mode with bowtie2. For the paired-end data I also rev/comp the R1 and combine both and map the in the SE mode. In both cases I do see stand specificity, but I think that this is not the proper way to do. bowtie2 handle them as unstranded reads and so they can be mapped on both strands and maybe cause higher background, or am I wrong?

So, is there anyone out there mapping bacterial dUTP RNASeq data with bowtie2? And what kind of flags do you set? Or what kind of program did you use despite bowtie2?
I tried all possible combinations with the --fr/--rf and --norc/--nofw flags and always one of the reads doesn't map at all and the other does not map in sense orientation.

Thanks a lot for advice or just discussion, so I don't have the feeling to spin round!

Sonja

Sonja,

See my reply up above.

[tokikake]

Dear kmcarr,
thanks for your reply; I have seen your answer but as they are flags for bowtie to distinguish between read orientation, I had the feeling it must be possible.

I tried tophat with the --library-type fr-firststrand option but it maps all the reads uniformely on both strands; no strand specifity is visible any more.

As I mentioned above I tried all possible options from bowtie. In the end, the only thing that worked nicely is to reverse complement R1 and map it (or both, R1 and R2, in case of PE data) with bowtie SE mode.

After lots of discussion in my lab and examination of dUTP libraries from different prokaryotic organisms and dates of library creation, I think the problem of my higher background is due to the library creation step, especially the degradation of the uridine-containing strand. Here we see big differences in the type of organisms (e.g. clostrida have nearly no background, whereas Proteobacteria does) and also in how often the kit has been used. I have data from the same organism where the libraries have been splitted with other preparations in between and I can see a higher background, when the kit has been freezed and thawed 3-5 times in between.

In the end we will try other vendors for dUTP Kits and make aliquots from the enzyme in the kit to prevent any activity loss. In the moment we use the NEBNext Ultra Kit from NEB.

Sonja