Multiple sequence threading: conditional gap placement

Preliminary work to improve on the gap placement in a novel multiple sequence threading method is presented here. Using the globin family, we construct measures for the assessment of gaps in a multiple sequence threading alignment based on the structural comparison of two of the proteins in the family. We take into account information from multiple sequence alignments on both the structure and sequence side of the problem. This work shows the parameterization of the problem allowing the foundation to optimize and test a gap placement weight or penalty. Four states were considered: deleted structure, inserted sequence, gap ends in structure, and broken ends in sequence. Each of these states was analyzed for exposure, occupancy and secondary structure. These measures enable us to gain insight into the placement of gaps in a multiple sequence threading alignment. We analyzed the most extreme violations of these properties and found in these cases that most secondary structures are broken by gaps. However, sequence inserts in structure were never found in deeply buried positions. Most end separations were 3-4 A in excess of the minimum 6 A, although some were larger. We show that the maximum amount of observed secondary structure found in inserts was about half of the predicted structure (typically 5 and 10%, respectively). A similar trend occurred with observed and predicted exposure in inserts. Our eventual aim is to devise a weight incorporating these measures of gap placement to further refine our algorithm for the threading of sequence on structure.