SiteHunter: Function prediction in CASP8

Method description

SiteHunter is an automated webserver for the prediction of protein-ligand and protein-DNA interactions through protein threading. SiteHunter comprises two method components: FINDSITE [1] that detects binding pockets for small molecules, and DBD-Hunter [2] that identifies DNA-binding sites. FINDSITE employs template identification, structure superimposition and binding site clustering as follows: First, for a given target sequence, structure templates are selected by three threading procedures: PROSPECTOR_3 [3], Sparks2 [4] and SP3 [5]. Subsequently, template structures bound to ligands are identified and superimposed onto the target protein structure using the structural alignment algorithm TM-align [6]. In CASP8, we used TASSER [7] models as the reference structures. After the superimposition, putative binding sites are inferred through the clustering of the template ligands, and the predicted sites are ranked according to the number of templates that share a common binding pocket. The second component of SiteHunter carries out the DNA-binding function prediction. This component utilizes a knowledge-based method, DBD-Hunter, which requires either the structure or sequence as the input for a given target. In CASP8, we applied the prediction procedure that requires only the sequence input. The method combines structural template threading and a statistical pair potential derived from known DNA-protein complexes. First, the target sequence is threaded against a template library composed of DNA-protein complex structures with the threading program PROSPECTOR_3 [3]. For template hits with a Z-score better than a threshold value, the statistical potential energy between the target protein and the template DNA is calculated by evaluating DNA-protein contacts. The templates are then ranked according to their interfacial energies. If the lowest interfacial energy is below an energy threshold, the target is predicted to be a DNA-binding protein. For predicted DNA-binding proteins, we further infer DNA-binding protein residues from their templates.

References

1. Brylinski,M. & Skolnick, J. (2008). A threading-based method (FINDSITE) for ligand-binding site prediction and functional annotation. Proc Natl Acad Sci U S A 105, 129-34.
2. Gao,M. & Skolnick,J. (2008). DBD-Hunter: a knowledge-based method for the prediction of DNA-protein interactions. Nucleic Acids Res 36, 3978-92.
3. Skolnick,J., Kihara,D. & Zhang,Y. (2004). Development and large scale benchmark testing of the PROSPECTOR_3 threading algorithm. Proteins 56, 502-18.
4. Zhou,H. & Zhou,Y. (2004). Single-body knowledge-based energy score combined with sequence-profile and secondary structure information for fold recognition. Proteins 55, 1005-13.
5. Zhou,H. & Zhou,Y. (2005). Fold recognition by combining sequence profiles derived from evolution and from depth-dependent structural alignment of fragments. Proteins 58, 321-8.
6. Zhang,Y. & Skolnick,J. (2005). TM-align: a protein structure alignment algorithm based on the TM-score. Nucleic Acids Res 33, 2302-9.
7. Zhang,Y., Arakaki,A.K. & Skolnick,J. (2005). TASSER: an automated method for the prediction of protein tertiary structures in CASP6. Proteins 61 Suppl 7, 91-8.

Preliminary assessment

• Target:
  CASP8 target ID as well as the PDB code and chain identifier of the experimental structure. The highest target-template sequence identity is given in parenthesis.
• Best pocket rank:
  FINDSITE rank for the best predicted ligand-binding pocket.
• Predicted binding residues:
  Protein residues predicted by FINDSITE to bind a ligand.
• Predicted binding pocket:
  Binding residues predicted by FINDSITE (surface) shown in the experimental structure (ribbon). A bound ligand as found in the crystal structure is presented as sticks.
• Native ligand:
  Small molecule complexed with the target protein in the experimental structure.
• Virtual screening:
  Top three compounds in the KEGG Compound library of >12,000 molecules ranked by FINDSITE. The Tanimoto coefficient to the native ligand is given in parenthesis.

Target	Best pocket rank	Predicted binding residues	Predicted binding pocket	Native ligand	Virtual screening
T0398 3d4oA (61.1%)	1	123, 129, 132, 136, 160-166, 184-186, 189, 203, 218-222, 224, 242-244, 264-268		TAR	1: C00006 (0.07) 2: C00005 (0.07) 3: C13051 (0.07)
T0422 3d8bA (31.6%)	1	85-87, 126-133, 259, 288-289, 292		ADP	1: C00206 (0.94) 2: C00131 (0.90) 3: C02509 (0.95)
T0426 3da2A (55.6%)	1	19-22, 115, 117, 119, 144, 154, 164, 166, 221-224		4MD	1: C07106 (0.54) 2: C14412 (0.46) 3: C07372 (0.51)
T0429 3db3A (18.2%)	1	79, 81, 83, 86, 88, 120, 130, 152, 154, 169-172		LIG	1: C01352 (0.12) 2: C00016 (0.12) 3: C00857 (0.14)
T0430 3dlzA (22.8%)	1	49-50, 57-58, 68-70, 116, 162-165, 212-213, 215, 225-226		AMP	1: C00939 (0.76) 2: C07195 (0.80) 3: C00212 (0.80)
T0431 3daxA (32.0%)	1	97-98, 112, 264, 267-269, 272, 343, 427-429		HEM	1: C00032 (0.87) 2: C06767 (0.38) 3: C15672 (0.79)
T0445 3daoA (25.8%)	1	10, 44-45, 125-126, 129, 149, 151, 153, 179-180, 182-185		1PE	1: C13741 (0.06) 2: C02591 (0.12) 3: C05001 (0.15)
T0445 3daoB (25.8%)	1	10, 44-45, 125-126, 129, 149, 151, 153, 179-180, 182-185		CIT	1: C13741 (0.08) 2: C02591 (0.16) 3: C05001 (0.18)
T0450 3da1A (41.6%)	1	24-30, 47-50, 52-62, 191-193, 227-230, 339, 341, 373-375, 377		FAD	1: C01469 (0.72) 2: C00008 (0.75) 3: C06743 (0.69)
T0477 3dkpA (29.5%)	1	1-2, 48-50, 53, 56, 75-81		ADP	1: C00131 (0.90) 2: C13742 (0.91) 3: C04380 (0.92)
T0483 3dlsA (29.6%)	1	32-33, 40, 53, 55, 92, 108-111, 116, 159, 162, 172-173, 312, 315		ADP	1: C16038 (0.27) 2: C01937 (0.28) 3: C11737 (0.22)
T0485 3dlcA (21.9%)	1	9, 49-53, 56, 71-73, 77, 99-101, 117-120, 123		SAM	1: C05692 (0.60) 2: C05691 (0.62) 3: C00021 (1.00)
T0490 3dmeA (26.4%)	1	10-16, 33-36, 40, 42-49, 171-173, 204-207, 212, 315, 317, 347-348, 352-354		FAD	1: C01469 (0.72) 2: C00008 (0.75) 3: C06743 (0.69)
T0490 3dmeA (26.4%)	1	10-16, 33-36, 40, 42-49, 171-173, 204-207, 212, 315, 317, 347-348, 352-354		TLA	1: C01469 (0.07) 2: C00008 (0.08) 3: C06743 (0.07)
T0494 2vx3A (29.1%)	1	62-64, 68, 70, 83, 85, 119, 135-138, 141, 188-189, 191, 203-204		D15	1: C00939 (0.19) 2: C07195 (0.17) 3: C00212 (0.17)
T0508 3douA (32.3%)	1	44-49, 51-52, 66-69, 82-84, 111-113, 151		SAM	1: C05692 (0.57) 2: C05691 (0.57) 3: C00021 (0.78)