File Changes for cline

v261_preview to v261_preview2 (2011-12-13 to 2011-12-20) v261

• src/hg/near/hgMapToGene/hgMapToGene.c
  • lines changed 43, context: html, text, full: html, text
    Resolving Redmine # 6093. The knownToRefSeq and knownToEnsembl tables contained cases where a small noncoding gene was overlapping a coding gene, often in the intron. To fix this, I did two things. First, I fixed a bug whereby if two candidate matches had the same score, the match to the shorter sequence was taken - even if both candidate matches had a score of zero. This generated a lot of false matches. Second, after discussing the situation with Jim, I added a requirement that as well as sharing some exonic bases, the two sequences must share a splice site unless they're both unspliced. This further decreases spurious matches between short ncRNAs and long coding genes