Help
StructAlign is a program for alignment of DNA-protein complexes' structures.
The program gets a pair of complexes of DNA double helix with proteins in PDB-format and optionaly names of protein chains and outputs an alignment of DNA chains corresponding to the best spatial fit of protein chains also in PDB-format. Moreover it returns score of the alignment, nucleotides with max measure and DNA alignment.
Score is a numeric quality of an alignment: the larger score - the better alignment.
Nucleotides with max measure are nucleotides, which superposition leads to the best alignment of protein structures.
Your results stores for a week.
How does it work
Firstly, for every nucleotide in both structures it associates a new coordinate system. This system is placed into phosphate group. X-axis is directed from P atom to C1' atom. Y-axis is a cross producr of X-axis and vector direected from P atom to a middle point between oxygens. Z axis is a vector product of X and Y axises.
Then for each nucleotide pair the program superimposes these structures and recalculate coordinates of protein atoms in new coordinate systems.
After that it calculates measure of protein similarity M as sum of positive subtractions of 4.5 and ρ, where ρ is a distance between corresponding Cα-atoms of structures A and B (corresponding atoms are found using bidirectional hits method):
$$M=\sum_{k,l}max \begin{cases} 0 \\ 4.5-\rho(k,l) \end{cases}$$Once all measures for all nucleotide pairs were obtained, the program builds similarity matrix (as shown below) and searchs best pair of complement diagonal, which corresponds to DNA fragments where proteins have the best superposition.
All what needs to be done at this point is to superimpose these fragments of DNA and generate PDB-file.