How To Split Bar Coded Solid Ngs Reads?

Question

[Ian]

Has anyone used Bioscope 1.2 to split barcoded and pooled SOLiD sequence reads? E.g. if i start with one .csfasta file containing six different barcodes I want to be able to split them into six different .csfasta files, each containing only one barcode.

I realise i can probably script this splitting process myself, but i would like to know how this can be achieved using Bioscope.

Thanks!

[Ian]

Thank you everyone for your comments. In the end i wrote my own script to split the barcoded reads. Basically it pairs the core read id (number_number_number) of the sequence and qualities scores with the barcode id.

I made the decision to allow missed colours (designated by '.' in the sequence and '-1' in the quality scores) and one wrong colour call. In both cases there has to be only one unambiguous match to one of the sixteen 5mer barcodes.

I am testing it at the moment, but if anyone is interested in taking a look i am happy to make it available.

Answer 1

[Istvan Albert]

Members of the Pugh Lab have developed a barcode splitter for the SOLiD platform. It is a perl script that you can get from:

http://bcc.bx.psu.edu/static/sep_tag_by_color_barcode-v1.1.pl

One neat feature is that it allows mismatches.

[Ian]

Unfortunately our reads have the barcodes already removed, but this is a very interesting script nonetheless.

[Istvan Albert]

You may be able to generate a new file with the barcodes added to the beginning of each read then the script would work.

[Ian]

OK that is probably worth trying.

[Istvan Albert]

One feature that I am exploring for the new version of BioStar is to integrate it with a source code version control system to allow people to share code that may be longer than a small snippet.

Answer 2

[Michael]

I hadn't heard about Bioscope, it seems to be a complex commercial suite by Applied Biosystems, but given there is the Fastx toolkit with the barcode splitter on Galaxy, it's maybe not necessary to use such a complex tool anyway.

Edit: Caveat: Fastx might not be able to process in colorspace, does someone know?

My recommendation: keep it simple, you possibly don't need to bother with such a complex (some people might call it bloated) tool as the one you are mentioning that might cost you license fees for trivial tasks.

Just as an aside point: In my opinion the vendors should focus more on their strength in producing rock-'solid' instruments and not on building software suits promising lofty "analysis without bioinformatics knowledge", because they cannot really offer you that, as they lack largely the programming and bioinformatics expertise themselves.

[Dave Lunt]

In case it helps, Galaxy URL: http://main.g2.bx.psu.edu/

[Ian]

Thanks i have noticed fastX before, but i don't think it handles SOLiD colour space reads...

[Michael]

Doesn't it? I didn't try it for that, ducumentation just states it handles fasta and fastq. Last resort: to convert to letter-space before splitting, but ofc you might want to align in color-space...

[Jorge Amigo]

I'm currently setting a local installation of Galaxy on our cluster, and I haven't found a way of splitting by barcodes using Galaxy only. as far as I understand the issue, we will have to pre-process our csfasta files with a custom barcode splitter such as the ones described here, and then feed the analysis software (galaxy or any other) with the splitted csfasta files.

@michael: BioScope is the free (until now) tool from Applied that claims to be very easy to install and use. but I can't really talk about its usability yet because we've waiting 2 month to have it installed ;)

[Jorge Amigo]

sorry, I just came through the fastx instructions on how to integrate their toolkit into Galaxy. FYI: http://hannonlab.cshl.edu/fastx_toolkit/galaxy.html

Answer 3

[Ian]

In the end i have written a script to answer my own question, but thanks for the previous answers!

Basically my script pairs the core read id (number_number_number) of the sequence and qualities scores with the barcode id. I made the decision to allow missed colours (designated by '.' in the sequence and '-1' in the quality scores) and one wrong colour call. In both cases there has to be only one unambiguous match to one of the sixteen 5mer barcodes.

I am testing it at the moment, but if anyone is interested in taking a look i am happy to make it available.

[Mikael Huss]

Hi Ian, could you share the script with me please? The email address is my username here @gmail.com.