Computational Analysis of the AK 092578 Locus

An RNAz-based (v1.0, [10]) search for structured RNAs within the 32k-region hg18.chr4:166,012,000166,043,999 yields only five ncRNA candidates. The input for RNAzwas generated with the mafsInRegion program, which is part of Kent’s UCSC-source-tree. For this locus, mafsInRegion reports 260 conserved sequence blocks within the UCSC 28-way vertebrate alignments. Further pre-processing of those 260 alignments was done using rnazWindows.pl, a part of the RNAz package, which selects up to six of the aligned sequences for screening by RNAz in such a way that their pairwise sequence similarity is optimized for the performance of RNAz. This is necessary because RNAz at present can evaluate alignments of no more than 6 species. The tool also prepares alignment windows of appropriate length and sequence composition matching the RNAz training scope. Actually only two of the five RNAz hits have an acceptable classification probability. The coding potential calculator (CPC, [8]) scores all RNAz hits as non-coding regions, but it votes for coding when it scores the complete AK092578 locus. Details of the RNAz hits are given in table 1 and predicted structures are available at http://www.bioinf.uni-leipzig.de/data/REVERSADE/.