User:ElNando888/WikiGetSat/Ideas/Sequences sorting in labs
< User:ElNando888 | WikiGetSat | Ideas
<p>Bottom-level, the discussion.</p><p>Compare with https://getsatisfaction.com/eternagame/topics/sequences_sorting_in_labs</p><p>One beautiful feature would be the possibility to wikify those contents.</p><p> </p><p>----</p><p>If I'm not mistaken, sequences in labs can be sorted, and the algorithm currently in use seems to be the <a rel="nofollow" href="http://en.wikipedia.org/wiki/Hamming_distance">Hamming distance</a>.</p><p>I'd like to propose a new sorting algorithm (which I dubbed “LDq9”), based on a <a rel="nofollow" href="http://en.wikipedia.org/wiki/Lee_distance">Lee distance</a> metric with a pseudo-alphabet of size 9 (or more). An example mapping would be:</p><p><code>A.G.<span>-</span>.<span>-</span>.U.C.<span>-</span>.<span>-</span>.<span>-</span><br />0.<span>1</span>.<span>2</span>.<span>3</span>.<span>4</span>.<span>5</span>.<span>6</span>.<span>7</span>.<span>8</span></code></p><p>Which would result in following specific distances:</p><p>A:G = 1 <br /> U:C = 1</p><p>G:C = 4 <br /> G:U = 3 <br /> A:U = 4 <br /> A:C = 4</p><p>The basic idea simply being that, changes within the same nucleotide classes (purines or pyrimidines) represent a short distance, while a change of class represent a larger jump.</p><p>I believe that this would give a somewhat better view of the similarity of sequences, specially in the context of switches.</p><p>-- ElNando888</p><p>----</p><p>Nice Idea.</p><p>-- jandersonlee</p><p>----</p><p>Hmm. Suppose I change a GC bond to CG. How does that get scored? And should it differ if it's a switch lab or not?</p><p>-- jandersonlee</p><p>----</p><p>GC to CG would be a +8 step. <br /> I don't think the metric should change between static labs and switch ones, but this idea of mine may prove making little difference with a simple Hamming in the case of static target structures. <br /> For switches, I'm almost convinced that this sorting would be a lot more accurate.</p><p>-- ElNando888</p><p>----</p><p>Worth a try if the coding isn't too much.</p><p>-- eternacac</p><p>----</p>