gsMapper and compiling unmapped reads

[yost]

I am very new to the gsMapper program and rather naive. I have used gsMapper to map our sequenced genome to a closely related reference genome. This has worked well. There was about 10% of the reads that were unmapped. A portion of the unmapped reads should represent novel sequences in our genome of interest. We have interest in identifying these novel sequences therefore I am wondering if there is an easy/straight forward approach to extract the unmapped reads and use them in a de novo assembly to get contigs of unique sequences that do not match the reference genome. We have about 60,000 unmapped reads hence I am looking for some kind of automated process.

Thanks in advance.
Chris

[kmcarr]

Chris,

Since you have access to the gsMapper I am going to assume that you have access to all of Roche's off instrument applications including the command line sff tools. Create a text file with the Accno identifiers (e.g. FZVLH7X02HJZVY) for all of your unmapped reads, one identifier per line. The gsMapper program creates a file as part of its output called 454ReadStatus.txt which lists the fate of read making it very easy to get a list of unmapped reads. You can then use this text file to extract a subset of the total reads from the original SFF file using the sfffile command:

% sfffile -i myaccnofile.txt -o myunmappedreads.sff myinputfile.sff

This command takes your original SFF file (myinputfile.sff) and outputs a new SFF file (myunmappedreads.sff) which contains only those reads listed in the text file created as described above (myuaccnofile.txt). You could then use this new SFF file containing only the unmapped reads as input to the Roche de novo assembler (gsAssembler).