m (Removed note about file extensions)
m (formatting tweak)
|Line 34:||Line 34:|
Revision as of 12:26, 7 March 2007
This page is aimed at any developers or coders interesting in understanding or extending the new Sequence Input/Output interface for BioPython, SeqIO.
If all goes well, the code will be available in the next release, probably BioPython 1.43.
Adding new file formats
Note: The details are still subject to change
To add support for reading a new file format, you must implement an iterator that expects a just file handle and returns SeqRecord objects. You may do this using:
- An iterator class subclassing something from Bio.SeqIO.Interfaces
- A generator function (using the yield keyword; suitable for simple formats)
- An ordinary function which returns an iterator. For example, you could build a list of SeqRecords and then turn it into an iterator using the iter() function.
You may accept additional optional arguments (an alphabet for example). However there must be one and only one required argument (the input file handle).
What you use as the SeqRecord's id, name and description will depend on the file format. Ideally you would use the accesion number for the id. This id should also be unique for each record (unless the records in the file are in themselves ambiguous).
When storing any annotations in the record's annotations dictionary follow the defacto standard laid down by the GenBank parser... I should try and document this more.
To add support for writing a new file format you should write a sub class of one of the writer objects in Bio.SeqIO.Interfaces
Then, the new format must be added to the relevant dictionary mappings in Bio/SeqIO/__init__.py so that the Bio.SeqIO.parse and Bio.SeqIO.write functions are aware of it.
Possible additional formats
There are existing parsers in BioPython for the following file formats, which could be integrated into Bio.SeqIO if appropriate.
NBRF / PIR format
Bio.NBRF has a Martel parser for this file format, which is similar to the FASTA format. It would need addition work to return SeqRecords. It might be easier to extend to reuse the Bio.SeqIO fasta code instead.
Can Bio.KEGG parse these files?
PHD sequencing files from PHRED
Bio.Sequencing.PHD has a Martel parser for this file format, also used by the tools PHRAP and CONSED.
MASE alignment format
Bio.IntelliGenetics seems to use Martel parse MASE format files into its own record object. It could be extended to return SeqRecord objects. See:
Bio.MEME has a parser for this file format, which at first glance looks like it could be treated like an alignment format.
Pairwise alignments from the BLAST suite could be turned into two SeqRecord objects with gapped sequences. Is this useful?
COMPASS pairwise alignment format
Bio.Compass can parse the pairwise alignments from COMPASS. The output is similar to BLAST in many ways. Again, is getting the results as SeqRecord objects useful?