From: Nomi H. <no...@us...> - 2005-12-15 00:30:10
|
Update of /cvsroot/gmod/apollo/doc In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv2158 Modified Files: todo fixed-bugs Log Message: Index: todo =================================================================== RCS file: /cvsroot/gmod/apollo/doc/todo,v retrieving revision 1.818 retrieving revision 1.819 diff -C2 -d -r1.818 -r1.819 *** todo 14 Dec 2005 00:27:34 -0000 1.818 --- todo 15 Dec 2005 00:30:01 -0000 1.819 *************** *** 5,9 **** --- 5,12 ---- ---------------------------------------------------- + Test restoring from serial file + ChadoXML issues + - Primary dbxref getting lost for annot dpp of dpp.chado! - It seems that the one-level annots don't come in with a <seqlen> field. Don't write <seqlen> for annots that don't have transcripts (and *************** *** 11,22 **** - featureprops need to have a rank (because there could be multiple values for one key)--is it ok to generate rank on the fly as we write out ! featureprops? ! - dbxref/feature_dbxref roundtripping issues: ! For the dbxref, the writer still tries to find the one that matches the ! feature's uniquename, but now if it can't find that, it uses the first ! xref. I think we need to add a method to tag the "real" xref. ! Also, currently all dbxrefs are also written out as feature_dbxrefs; we ! need to make sure that we only write feature_dbxrefs that came in as ! feature_dbxrefs. - Strange behavior displaying dbxrefs in AIE. If you invoke AIE on a transcript (e.g. skpA), all the dbxrefs for ALL the transcripts, --- 14,23 ---- - featureprops need to have a rank (because there could be multiple values for one key)--is it ok to generate rank on the fly as we write out ! featureprops? (Note that if featureprops didn't come in in rank order, ! their ranks will change--e.g. dpp.chado has featureprops with type ! "citation" where the one with rank=1 is before the one with rank=0, so ! they end up renumbered when you save.) ! - Roundtrip ranks for comments. Would it be helpful to roundtrip the cv ! name for these? - Strange behavior displaying dbxrefs in AIE. If you invoke AIE on a transcript (e.g. skpA), all the dbxrefs for ALL the transcripts, *************** *** 29,37 **** value is in fact an int. [any other fields we need to integrity-check?] - Although flat fields like <is_obsolete> are now roundtripped properly ! (as properties inside Apollo) when they are attached to features, the ! <is_obsolete> and <is_relationshiptype> fields are now also appearing ! inside type_id records, where they are NOT currently being preserved. ! [Andy and Pinglei say: ok to leave them out. So be aware that fields ! inside type_id records will not be roundtripped.] - Doesn't yet handle macros! - Trans-spliced transcripts not really dealt with correctly--I don't --- 30,38 ---- value is in fact an int. [any other fields we need to integrity-check?] - Although flat fields like <is_obsolete> are now roundtripped properly ! (as properties inside Apollo) when they are attached to features, in some ! ChadoXML data the <is_obsolete> and <is_relationshiptype> fields are now ! also appearing inside type_id records, where they are NOT currently being ! preserved. [Andy and Pinglei say: ok to leave them out. So be aware ! that fields inside type_id records will not be roundtripped.] - Doesn't yet handle macros! - Trans-spliced transcripts not really dealt with correctly--I don't *************** *** 72,75 **** --- 73,85 ---- One-level annots: ----------------- + - It used to be that if you rubberbanded an region in the annotation + thereby selecting a number of transposable_element_insertion sites for + example, and then chose the 'delete feature' option from the right mouse + popup then all the features of that type that were selected would be + deleted. Now it seems that even if multiple features of the same type + are selected only a single feature is deleted at a time. This only + appears to apply to the single level features as you can delete multiple + transcripts by selecting them all and choosing to delete. Is this + intentional. - Desired EDE behavior for first/second tier annots (according to Lynn): open EDE on a second tier annot -> show all annots (as now) *************** *** 220,223 **** --- 230,244 ---- message (and non-graceful recovery) + Data interconversion issue: + Flat fields in the Chado XML input are stored as properties inside Apollo + prepended with the FIELD_LABEL string, e.g. + <is_obsolete>0</is_obsolete> -> addProperty(feature, "field:is_obsolete", + "0") + The ChadoXML writer knows to remove the FIELD_LABEL string, and I also + put that in the GAME writer, but other data adapter writers (including + any new ones that might get written) don't know to do this. + This may become even more of a problem as other special prefixes are + added to property keys for FlyBase purposes. + Warn on edit: - warnOnEdit sometimes doesn't work (lets you edit second-tier annots w/o Index: fixed-bugs =================================================================== RCS file: /cvsroot/gmod/apollo/doc/fixed-bugs,v retrieving revision 1.498 retrieving revision 1.499 diff -C2 -d -r1.498 -r1.499 *** fixed-bugs 14 Dec 2005 00:27:34 -0000 1.498 --- fixed-bugs 15 Dec 2005 00:30:01 -0000 1.499 *************** *** 1862,1863 **** --- 1862,1872 ---- ---------^^^----------Internal Release 1.6.2 (12/13/05)----------^^^--------- + + dbxrefs and feature_dbxrefs now roundtrip properly (MOSTLY). They are + tagged (as primary and secondary dbxrefs) as they are read in, and then + they are written out as appropriate. The writer assumes that there will + be only one primary dbxref (although this would be easy to change, and + the reader will record multiple primary dbxrefs if they appear in the + input). It is also ok if the primary dbxref for a feature does not + appear as a feature_dbxref (in the past, Apollo assumed that the primary + dbxref was also a feature_dbxref, but that's not always true). |