Fig. 1

Flowchart of the method as implemented in MOMA. a Example of structure of MarA (PDB code 1bl0) and the matrix representation of its folding pattern. The relative orientation of any two secondary structural elements (for example, A4 and A6 helices) is specified by the angle (w) between the vectors along their axes (left bottom of the matrix). This is recorded only for those SSE pairs found in close proximity (d < D), as measured by the distance (d) between midpoints of the vectors (upper right of the matrix). b These matrices are built for the query (1BL0 chain A) and the target (1AIH chain A) structures. After that, row-wise matrices containing all possible SSE pairs in each structure are also built. Query and target proteins render matrices of [M, M-1] and [N, N-1] pairs, where M and N correspond to the total number of SSEs found in the query and target structures, respectively. c A first step of global or semi-global dynamic programming (DP) algorithm is executed to build DP matrices for each query row against each target row, thus generating a total of MxN DP matrices. In this step, scoring rules and restraints based on angular and distance information of all SSE pairs in each structure are used (see Methods for details). From each DP matrix, only the maximum score value is selected and recorded into a new scoring matrix that is going to be used in a second and final step of a dynamic programming algorithm. In the case of a global alignment, this value is obtained from the bottom right cell of the DP matrix. In the case of the semi-global alignment, this value is obtained from the most right column or the most bottom row of the DP matrix. d A local dynamic programming algorithm and the previously built scoring matrix are now used to align the secondary structure elements of the query and the target structures. e Unaligned SSE elements from the query and target structures are removed from the initial 2D matrices, thus rendering two matrices of identical dimensions, which can now be compared directly. A delta sub-matrix is built and from it a global matching score calculated (see Methods for details). f Finally, a new algorithm (*) is used to infer the list of all incompatible rigid local matches (blocks), which are independently superposed with the Kabsch algorithm. In this particular and simple example only one local match or block is found. Details of the algorithm for finding local matching blocks are provided as Supplemental Material (Additional file 1: Figure S1). The resulting superposition is represented with aligned elements in red (query) and orange (target). Residues not aligned are displayed in grey color