Unmapped ratio very high on mouse genome

[le.nono]

Hi,
My problem regards RNA-Seq data. I've downloaded public data (SAGE libs w/ 6 different samples from mouse liver ) to analyse using ArrayStudio. When I try to map them on the B38 mus musculus genom I have an unmapped read % of approximatly 95 % on all the samples!!! Quality scores are correct around 40 read length is correct (35 bp) but the base distrib QC is just very heterogenous, I don't understand why... this the first time I work on mouse data.Does anybody shared the same problem or have an idea please regarding the mapping and/or the base distrib?

Thanks, LN

[GenoMax]

You may want to do a FastQC run on the data first to check on the quality. The data you downloaded may be raw and you may need to trim/clean the data before doing analysis/alignments.

[le.nono]

They say there is a 16 bp adaptor on each read, but my reads are at the correct length 35 bp on the QC. Do I really need to trim them?

[le.nono]

Quality score histogrammes look very good for each sample.

[GenoMax]

They say there is a 16 bp adaptor on each read, but my reads are at the correct length 35 bp on the QC. Do I really need to trim them?

Is this supposed to be an "inline" adapter that is part of the actual sequence? Are you able to tell by looking at the reads?

[le.nono]

I don't think so the reads are very short. What do you have in mind?

[GenoMax]

Can you post a FastQC (or which ever kind of QC you used) graph of the base distribution?

I was thinking that one way you would get 95% of reads unmapped is if the barcodes/adapter were still present in the reads (inline). Do you know if they have already been removed?

[le.nono]

no I don't have this information.

[le.nono]

Maybe a better quality and size one.

http://imageshack.us/photo/my-images/844/q6c.png/

[GenoMax]

All the sequences appear to be starting with exactly the same 4 nucleotides (GCCA). Is that a barcode?

[SNPsaurus]

Are you able to map other SAGE data with your pipeline? Maybe it is not set up for such short tags.

The 4bp starting sequence is the cut site, right?

Also, are these ditags of 16 bp? Those would not map unless you split them first.

[le.nono]

I m gonna try to trim those 4bp first map the reads. I definitely need further informations on the reads... I dont much about those 16 bp adapter its just written in the abstract coming with the data. Do you say that what is display on the base distrib histrogrammes are ditags of 16 bp?

[SNPsaurus]

If it is SAGE data, then you should look here for an overview of the method:

http://www.scq.ubc.ca/painless-gene-expression-profiling-sage-serial-analysis-of-gene-expression/

It is an older method meant to increase the sampling of transcripts with Sanger sequencing. There are some mouse mapping tools here:

http://www.webcitation.org/getfile?fileid=034304f3130c539c0d33e68f15243e9258904b3a

But I suspect you'll want to find some newer RNA-Seq data that isn't SAGE based and you'll find it easier to go forward.

[le.nono]

Ok it s becoming clearer now. I really need these data I use so i m gonna stick to them even if it s harder. I m gonna try to look for in the literature some RNA Seq with SAGE preps I think its been done before what do you think?