EMBOSS: C2 Output Formats

Any format can be used for output, but some formats simply write a sequence and have no problem if the output file is closed at the end of the program. Others, e.g. MSF, store multiple sequences and write them only when the file is closed by a FileClose call.

This means that all applications should use the ajSeqWriceClose call to make sure sequence output is flushed.

Simple output formats:

• GCG. EMBOSS is using code from readseq for the checksum. GCG have said that the algorithm is in the public domain.
• FASTA (also called PEARSON) with accession number and description
• NCBI (FASTA with NCBI formatted id and accession)
• PIR
• EMBL
• SWISS
• GENBANK
• ASN.1 (readseq's shortened version - perhaps not available as an input format)
• FITCH
• IG
• NBRF (also called PIR)
• CODATA
• EXPERIMENT also known as STADEN
• TEXT
• STRIDER
• DEBUG (was TRACE) a sequence object detailed report mainly for debugging purposes.

Multiple output formats:

• MSF
• CLUSTAL or ALN
• PHYLIP

The aim is to include all formats that EMBOSS accepts for input plus some formats intended for output only.

Issues:

1. Some formats require data which may be unavailable. For example, TEXT input from standard input has no ID, but EMBL output requires an ID. This could default or could be provided on the command line as "-sentry" for use on output.
2. The default "id:" can be set to EMBOSS or EMBOSS_00n for output formats that need one.
3. The default accession number can be any deleted or never used accession number, to be agreed with the EBI. Obvious candidates are M12345 and X00000. It is not clear whether any format absolutely requires an accession number.
4. The "id"can be specified in an output USA. There is currently no way to specify an accession number this way, unless we try NCBI syntax of "id|accnum" but this seems more trouble than it is worth.

Other points

• "debug" format is a report of the internal sequence object structure. It is used for debugging and is not intended as a new sequence format.
• The output can be sent to a file or to stdout.
• Output specifications should use USA syntax to specify the format and filename. Output specs now allow "format::file" We tried to allow "format::file:id" but this has the problem that "format:file" is valid, with "format" as the filename and "file" as the entry ID. This is a very easy mistake. The "id" is therefore not allowed on output, and the USA can be just "format::file" (recommended) or "format:file" (to rescue careless users)