Processing Barcoded Paired-end Reads

[anthonycychiu]

I've never processed RNA-seq reads (only done bowtie for ChIP-Seq processing) so I wanted to clarify some things. I found a dataset off GEO where the paired end reads have been concatenated into a single 282 read. I'm wondering what the best way to process it is?

Is the best way to split it in half into read1 and read2, and then trim adapters from them before mapping with Tophat. However I don't know if thats correct; is it better to map it unsplit?

Also, for one thing I don't know what the barcode sequence is for some of the datasets are (FASTQC didn't give me the barcdode for 5 of the 8 samples). Has anyone tried trimming adapters using TTTTTTNNNNNATTTTTTT, where N refers to the barcode sequence in FASTX-CLIPPER?

Thanks.

[Brian Bushnell]

If the merging process was done correctly, the adapters will already have been removed. If it was not done correctly, you shouldn't use that data, but get the original data instead. And yes, you should map them as-is rather than trying to split them.

P.S. And in my testing, BBMap is both faster and more accurate for RNA-seq than Tophat.

[GenoMax]

@anthonycychiu: Have you tried fastq-dump with the following option to get the two separate files for R1 and R2?

--split-files Dump each read into separate file.Files
will receive suffix corresponding to read
number

[Brian Bushnell]

@anthonycychiu: Have you tried fastq-dump with the following option to get the two separate files for R1 and R2?

Ah - if this was a fastq that you got by dumping an SRA archive, my answer might be totally wrong and should be ignored.