EMBOSS: Project Meeting (Mar 30th 2000)
The rounding problem in needle has been tracked down, cleaned up, wiped out and checked in by Alan. The cause was a cumulative error in subtraction that floating point rounding was still not curing.
Ian has fixed problems with axis labelling in dottup and dotmatcher.
Alan has modified restrict. There is a new scanning routine for restriction sites, with fixes for begin and end positions and mincuts and maxcuts. Palindromic cutters are marked only on the top strand in all cases.
revseq has a bug in converting the W and S ambiguity codes. They should be unchanged when a seqeunce is complemented (S is G or C, W is A or T).
Sequence ranges need to be modified to make them easier to use. Peter proposed defining an associated sequence, with "all" as the default range (and "start" and "end" as aliases for the first and last base). This would check the sequence begin and end position in ACD processing, and pass on the adjusted sequence range to applications.
Peter proposed changing "-sentry" to "-sid" to help reduce ambiguous qualifier names, though "-send" would still clash with "-sequence" in many applications.
Output files are created even if they are not needed by the application. A new attribute is needed to make them optional, with an associated qualifier to force creation of a new file for scripts or interfaces that will check for the existence of the output file automatically.
Peter is adding USA generation. For some cases, the USA will need to be a filename and offset. The provisional syntax is "filename%offset"
Val is having problems with long sequences in supermatcher.
Peter suggested increasing the default work size on supermatcher and other word based programs to 10 as small values lead to long run times on large sequences and many users are simply using the default values.
Val's tutorial needs to be added to the web pages, as a link to the online version at HGMP, and also included in the distribution.
Meeting on Monday 10th April, 11:00am, usual place.