Comments on: Heads or tails and reliable alignments http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/ Computer science, bioinformatics, genetics, and everything in between Fri, 19 Aug 2011 16:35:39 +0200 hourly 1 http://wordpress.org/?v=3.0.4 By: Reed A. Cartwright http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/comment-page-1/#comment-1002 Reed A. Cartwright Sun, 23 Mar 2008 21:03:09 +0000 http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/#comment-1002 I whole heartedly agree. I whole heartedly agree.

]]> By: Thomas Mailund http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/comment-page-1/#comment-996 Thomas Mailund Sun, 23 Mar 2008 15:43:11 +0000 http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/#comment-996 My feeling is that we should ignore the alignment altogether (in most cases) and focus on the analysis we <em>really</em> want to do. In most cases, we use the alignment as a step towards some other analysis, and I would be much happier to consider the alignment missing data and (to the extend possible) just integrate it out of the equation. My feeling is that we should ignore the alignment altogether (in most cases) and focus on the analysis we really want to do. In most cases, we use the alignment as a step towards some other analysis, and I would be much happier to consider the alignment missing data and (to the extend possible) just integrate it out of the equation.

]]> By: Reed A. Cartwright http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/comment-page-1/#comment-990 Reed A. Cartwright Sun, 23 Mar 2008 14:12:16 +0000 http://www.mailund.dk/index.php/2008/03/23/heads-or-tails-and-reliable-alignments/#comment-990 In my alignment program, Ngila, I made choices about how to break ties to give a "pretty" alignment. IIRC, I chose to shift gaps as far right as they can go, which means reversing the sequences will give you a different alignment. However, I'm not a big fan of relying too much on "optimal" alignments. I'd rather see more approaches treat alignments are hidden data and try to optimize around their set. In my alignment program, Ngila, I made choices about how to break ties to give a “pretty” alignment. IIRC, I chose to shift gaps as far right as they can go, which means reversing the sequences will give you a different alignment.

However, I’m not a big fan of relying too much on “optimal” alignments. I’d rather see more approaches treat alignments are hidden data and try to optimize around their set.

]]>