ProSMoS server: a pattern-based search using interaction matrix representation of protein structures

Abstract

Assessing structural similarity and defining common regions through comparison of protein spatial structures is an important task in functional and evolutionary studies of proteins. There are many servers that compare structures and define sub-structures in common between proteins through superposition and closeness of either coordinates or contacts. However, a natural way to analyze a structure for experts working on structure classification is to look for specific three-dimensional (3D) motifs and patterns instead of finding common features in two proteins. Such motifs can be described by the architecture and topology of major secondary structural elements (SSEs) without consideration of subtle differences in 3D coordinates. Despite the importance of motif-based structure searches, currently there is a shortage of servers to perform this task. Widely known TOPS does not fully address this problem, as it finds only topological match but does not take into account other important spatial properties, such as interactions and chirality. Here, we implemented our approach to protein structure pattern search (ProSMoS) as a web-server. ProSMoS converts 3D structure into an interaction matrix representation including the SSE types, handednesses of connections between SSEs, coordinates of SSE starts and ends, types of interactions between SSEs and beta-sheet definitions. For a user-defined structure pattern, ProSMoS lists all structures from a database that contain this pattern. ProSMoS server will be of interest to structural biologists who would like to analyze very general and distant structural similarities. The ProSMoS web server is available at: http://prodata.swmed.edu/ProSMoS/.

Figure 1.

(A) Interaction types used to define a query matrix. Yellow arrows and blue cylinders denote β-strands and α-helices, respectively. Smaller arrow inside each cylinder indicates the direction of a helix. Solid line between two strands shows H-bonding. Dashed line between SSEs means interaction other than H-bonding. ‘c’ and ‘t’ refer to the presence of interaction between hydrogen-bonded parallel and antiparallel β-strands, respectively. ‘u’ and ‘v’ mean there is an interaction between parallel helix-helix or helix-strand pair and the angle between the pair is 0 ≤ φ < 85° and 95° ≤ φ < 180°, respectively. ‘C’ is a union of {c, u}, i.e. interaction present whether through H-bond or not, but the angle between elements is 0 ≤ φ < 85°. ‘T’ is a union of {t, v}, i.e. interaction present whether through H-bonds or not, but the angle between elements is 95 ≤ φ < 180°. ‘−’ is used to indicate the absence of any interaction. ‘N’ means that there is an interaction between two SSEs, but the angle between them is 85° ≤ φ < 95°. ‘X’ means any relationship, i.e. interaction present or absent: a union of {C,T,N, −}. ‘x’ means any interaction, i.e. a union {C,T,N}. (B) Ferredoxin-like fold diagram. Ferredoxin fold is an α + β t-lawoyer sandwich with the secondary structure order βαββαβ. Four strands form an antiparallel β-sheet, which is covered by two α-helices on one side. (C) Query meta-matrix for the ferredoxin-like fold pattern.

Figure 2.

Representative ferredoxin-like motif hits. β-Strands shown in yellow and α-helices shown in cyan. The first residue of the motif is colored green and the first residue of each SSE is colored purple. (A) The best-scoring motif hit, pdb id 1sc6, chain D, D-3-phosphoglycerate dehydrogenase, residue ranges 337–346, 347–362, 364–373, 375–384, 386–400, 402–410. (B) The weakest-scoring motif hit (#8299), pdb id 3b8m, chain B, bacterial polysaccharide co-polymerase, residue ranges 64–74, 99–114, 152–157, 171–179, 180–251, 320–324. (C) The hit #482, pdb id 2az1, chain E, nucleoside diphosphate kinase, residue ranges 6–13, 20–33, 34–43, 73–81, 83–93, 117–121. (D) The hit #4750. This motif overlaps with #482 (C) over 5 SSEs, 6–13, 20–33, 34–43, 73–81, 104–110, 117–121, with the second helix being distinct (residue range 104–110, colored red). Since this helix is shorter than in (C), this motif scores worse. (C) and (D) illustrate the possibility of overlapping motifs being found by the server.

Figure 3.

Response time of ProSMoS Server for eight wide-spread structure patterns in complete PDB database. The response time (min) is recorded from the submission of a job to the receipt of result for each structure pattern.

Figure 4.

Comparison between ProSMoS and Dali. The red curve is the average precision of ProSMoS searches of eight common structure patterns. The dark blue curve is the average recall. The x-axis is the Dali Z-score used to define ‘true’ matches.