Welcome to phosfinder!

Overview of the method

Many important chemical reactions and molecular interactions that occur in the cell involve ligands containing the phosphate group. More than half of known proteins has been shown to interact with a phosphate group (1). Several of these proteins are involved in essential pathways and their malfunction leads to severe diseases and other abnormalities in humans (2,3). Moreover the affinity for the phosphate group is essential in nucleotide recognition and nucleotide-containing ligands were the earliest cofactors bound to proteins (4). The possibility to characterize a protein for its ability to interact with a phosphate, or a phosphate-containing ligand, is therefore of paramount importance. Pfinder (5) is a new comparative method for the identification of phosphate binding sites (PbSs) on a protein structure. The method performs a local structural comparison between the query protein and a data set of PbSs (6) thus identifying groups of amino acids, which display a similarity to known binding motifs. Pfinder then evaluates all the candidate predictions using several geometric criteria and a sequence conservation score to select the final solutions. Here, we present Phosfinder, the web server interface of Pfinder that allows the user to predict phosphate binding sites in a protein structure. Phosfinder uses the Pfinder core to predict the binding sites. Finally the web server lists all the predicted phosphate binding sites ranked according to a conservation score (calculated whenever the query protein structure can be associated to a PFAM multiple alignment).

Web server pipeline

The query protein can be submitted as a four-letter PDB code or uploaded as a PDB-formatted file:

- Many controls on the format and quality check are performed on the structure file and on the values and parameters given as input. The input values must be supplied as specified in the "usage" page, while the uploaded structure must contain the coordinates section of a PDB structure.
The query protein structure is compared to a set of known phosphate binding motifs using a structural comparison algorithm (7,8). During this phase, two parameters are considered:
- only structural matches with a Root Mean Square Deviation (R.M.S.D.) value lower than the threshold, specified in the search form, will be kept.
- only amino acids substitutions with a score higher than the BLOSUM62 threshold, specified in the search form, are allowed in a structural match.
Whenever a known phosphate binding motif matches with the query protein residues, the phosphate group bound by the known motif is roto-translated and added onto the query protein structure. The predicted binding sites are then evaluated and clustered as follows:
- Predicted phosphate binding sites located inside the solvent-exclude surface of the protein are discarded.
- The predictions are clustered using a hierarchical clustering (centroid-linkage with a threshold of 2.0 Å)
- A conservation score (ranging from 0 to 100) is given, when a corresponding high-quality PFAM multiple alignment is available, to each prediction. This score is calculated as the average of the conservation value of each residue forming the predicted binding motif. This value ranges from 0 to 100 and is calculated as described in (5).
- Finally the predictions are ranked according to the conservation score. Whenever the conservation score can not be calculated (i.e. no PFAM is available), the predictions are reported at the bottom of the chart and are associated to the no_cons symbol.


1. Hirsch KH, Fischer FR, Diederich F. Phosphate recognition in structural biology. Angewandte. Chemie Int. Edn 2006;46:338-352.
2. Traxler P, Furet P. Strategies toward the design of novel and selective tyrosine kinase inhibitors. Pharmacol. Ther. 1999;82:195-206.
3. Gitlin JD. Wilson disease. Gastroenterology 2003;125:1868-1877.
4. Ji HF, Kong DX, Shen L, Chen LL, Ma BG, Zhang HY. Distribution patterns of small-molecule ligands in the protein universe and implications for origin of life and drug discovery. Genome Biol. 2007;8:R176.
5. Parca,L, Gherardinin,PF, Helmer-Citterich,M, Ausiello,G. Phosphate binding sites identification in protein structures. Nucleic Acids Res. 2010; in press (doi: 10.1093/nar/gkq987).
6. Ausiello G, Gherardini PF, Gatti E, Incani O, Helmer-Citterich M. Structural motifs recurring in different folds recognize the same ligand fragments. BMC Bioinformatics 2009;10:182.
7. Ausiello G, Via A, Helmer-Citterich M. Query3d: a new method for high-throughput analysis of functional residues in protein structures. BMC Bioinformatics 2005;6 Suppl. 4:S5.
8. Gherardini PF, Ausiello G, Helmer-Citterich M. Superpose3D: a local structural comparison program that allows for user-defined structure representations. PLoS One 2010;5:e11988.