Commits for cline

v260_base to v261_preview (2011-12-06 to 2011-12-13) v261

• Making two refinements on how we use the Rfam alignments. First, remove any duplicate hits. In some cases, Rfam has duplicate hits from closely-related models (e.g. one model being a subtype of another), and in these cases we only want one of them. Second, before combining Rfam (and trnas) with the txWalk output, weed out any alignment that overlaps a txWalk alignment. In most cases, these will be duplicates
  • src/hg/makeDb/doc/ucscGenes/hg19.ucscGenes13.csh - lines changed 37, context: html, text, full: html, text
• When parsing the Rfam data, skip over comment lines
  • src/hg/makeDb/doc/hg19.txt - lines changed 14, context: html, text, full: html, text
• When parsing the Rfam data, skip over comment lines
  • src/hg/makeDb/doc/hg19.txt - lines changed 13, context: html, text, full: html, text
• Fixing Redmine Bug #3497, in which the wrong UniProt protein is mapped to a specific gene. The code read through the PSL alignments of UniProt sequences to the gene to pick the best one. In picking the best one, it computed a heuristic score with credits for the matching positions and penalties for the various types of alignment errors. But while computing the score, which was a double, unsigned ints were effectively multiplied by negative numbers without type conversion, resulting in a quantity close to MAX_INT. The net effect of this was that poor-quality alignments were receiving massive scores, and were being chosen as the 'best' alignments
  • src/hg/txCds/txCdsRedoUniprotPicks/txCdsRedoUniprotPicks.c - lines changed 7, context: html, text, full: html, text