Improvements in experimental techniques increasingly provide structural data relating to protein-protein relationships. for template-based docking. We test the specificity of these nonredundant unique interface structures by getting protein pairs which have multiple binding sites. We suggest that residues with more than 40% relative accessible surface area should be considered as surface residues in template-based docking studies. This comprehensive study of protein interface constructions can serve as a source for the community. The dataset can be utilized at http://prism.ccbb.ku.edu.tr/piface. Intro Proteins actually interact with each other through their binding sites. Some proteins interact with their partners simultaneously using different connection sites some interact with their partners via the same connection site at different times and some appear to interact with only one protein [1]. How can many different proteins use the same binding interface and how can a single protein bind many different proteins at the same time are key questions that emerge GDC-0449 in structurally-enriched protein-protein connection networks and rules. GDC-0449 Within the platform of the general factors that carry on these intriguing questions is the scenery of residue conformations particularly of key residues making multiple and simultaneous relationships possible [2]-[7]. While a vast number of protein-protein relationships can take place there is a limited quantity of specific binding site conformations through which proteins can bind [8]-[10]. Studies of interfaces can be illuminating: they can address questions such as whether preferences of specific amino acids in certain positions can help binding site prediction and on a different level how some proteins can bind many different proteins using the same binding site conformations. Since binding and folding are related events they may also help understand hierarchical protein folding [11]. Obtaining a set of unique interface constructions can be particularly useful in template-based docking [12] [13]. We previously showed that template centered docking can GDC-0449 be fast and accurate if there exists a good set of template interfaces [14] [15]. Kundrotas et al. posited that unique interface constructions can serve as themes to model nearly all complexes of GDC-0449 structurally characterized proteins and that the existing interfaces already can achieve this goal [13]. Kundrotas and Vakser showed that structural similarities of the interfaces have the greatest influence in template-based docking [16]. The main step to accomplish GDC-0449 a unique interface set is definitely to flag related interfaces. Comparisons of two protein interfaces can detect similarities in amino acid sequences of protein GDC-0449 interfaces (sequence alignment) or similarities in 3D coordinates of amino acid positions in the proteins (non-sequential structural alignment). Protein interface clustering can be done in three different ways: using only sequential or only structural alignment scores of all protein interfaces or a cross strategy which includes both sequential and structural positioning scores of protein interfaces. The PFAM [17] and SCOP [18] databases are commonly utilized for classification by sequence and structural alignments respectively. Previous studies aiming to investigate binding properties showed that protein interfaces Rabbit Polyclonal to GATA6. can be classified by their sequence similarities [8] [19] or in other words proteins with related sequences often interact in related ways [20] [21]. However it has also been suggested that interactions can only become reliably inferred for close homologs [20] [22]. To decrease the computational cost studies have also used a hybrid strategy of both sequence and structural assessment and to increase the reliability of the classification others used structural alignment of protein interfaces [23]-[27]. Bordner and Gorin clustered biologically relevant interfaces having a hybrid strategy to provide a reliable catalog of evolutionary conserved protein-protein interfaces having a diverse set of properties [28]. Detecting evolutionarily related proteins via structural similarities is more reliable than via sequence similarities since structure is more conserved. Sequence centered methods are easy to derive and computationally cheap so sequence centered methods are generally the starting.