MOTIVATION: Accurate prediction of protein stability changes upon single-site variations (ΔΔG) is important for protein design, as well as for our understanding of the mechanisms of genetic diseases. The performance of high-throughput computational methods to this end is evaluated mostly based on the Pearson correlation coefficient between predicted and observed data, assuming that the upper bound would be 1 (perfect correlation). However, the performance of these predictors can be limited by the distribution and noise of the experimental data. Here we estimate, for the first time, a theoretical upper-bound to the ΔΔG prediction performances imposed by the intrinsic structure of currently available ΔΔG data. RESULTS: Given a set of measured ΔΔG protein variations, the theoretically "best predictor" is estimated based on its similarity to another set of experimentally determined ΔΔG values. We investigate the correlation between pairs of measured ΔΔG variations, where one is used as a predictor for the other. We analytically derive an upper bound to the Pearson correlation as a function of the noise and distribution of the ΔΔG data. We also evaluate the available datasets to highlight the effect of the noise in conjunction with ΔΔG distribution. We conclude that the upper bound is a function of both uncertainty and spread of the ΔΔG values, and that with current data the best performance should be between 0.7 and 0.8, depending on the dataset used; higher Pearson correlations might be indicative of overtraining. It also follows that comparisons of predictors using different datasets are inherently misleading. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: Predicting the effect of single point variations on protein stability constitutes a crucial step toward understanding the relationship between protein structure and function. To this end, several methods have been developed to predict changes in the Gibbs free energy of unfolding (∆∆G) between wild type and variant proteins, using sequence and structure information. Most of the available methods however do not exhibit the anti-symmetric prediction property, which guarantees that the predicted ∆∆G value for a variation is the exact opposite of that predicted for the reverse variation, i.e., ∆∆G(A → B) = -∆∆G(B → A), where A and B are amino acids. RESULTS: Here we introduce simple anti-symmetric features, based on evolutionary information, which are combined to define an untrained method, DDGun (DDG untrained). DDGun is a simple approach based on evolutionary information that predicts the ∆∆G for single and multiple variations from sequence and structure information (DDGun3D). Our method achieves remarkable performance without any training on the experimental datasets, reaching Pearson correlation coefficients between predicted and measured ∆∆G values of ~ 0.5 and ~ 0.4 for single and multiple site variations, respectively. Surprisingly, DDGun performances are comparable with those of state of the art methods. DDGun also naturally predicts multiple site variations, thereby defining a benchmark method for both single site and multiple site predictors. DDGun is anti-symmetric by construction predicting the value of the ∆∆G of a reciprocal variation as almost equal (depending on the sequence profile) to -∆∆G of the direct variation. This is a valuable property that is missing in the majority of the methods. CONCLUSIONS: Evolutionary information alone combined in an untrained method can achieve remarkably high performances in the prediction of ∆∆G upon protein mutation. Non-trained approaches like DDGun represent a valid benchmark both for scoring the predictive power of the individual features and for assessing the learning capability of supervised methods.

Recent insights suggest that non-specific and/or promiscuous enzymes are common and active across life. Understanding the role of such enzymes is an important open question in biology. Here we develop a genome-wide method, PROPER, that uses a permissive PSI-BLAST approach to predict promiscuous activities of metabolic genes. Enzyme promiscuity is typically studied experimentally using multicopy suppression, in which over-expression of a promiscuous 'replacer' gene rescues lethality caused by inactivation of a 'target' gene. We use PROPER to predict multicopy suppression in Escherichia coli, achieving highly significant overlap with published cases (hypergeometric p = 4.4e-13). We then validate three novel predicted target-replacer gene pairs in new multicopy suppression experiments. We next go beyond PROPER and develop a network-based approach, GEM-PROPER, that integrates PROPER with genome-scale metabolic modeling to predict promiscuous replacements via alternative metabolic pathways. GEM-PROPER predicts a new indirect replacer (thiG) for an essential enzyme (pdxB) in production of pyridoxal 5'-phosphate (the active form of Vitamin B6), which we validate experimentally via multicopy suppression. We perform a structural analysis of thiG to determine its potential promiscuous active site, which we validate experimentally by inactivating the pertaining residues and showing a loss of replacer activity. Thus, this study is a successful example where a computational investigation leads to a network-based identification of an indirect promiscuous replacement of a key metabolic enzyme, which would have been extremely difficult to identify directly.

Conformational changes mediate important protein functions, such as opening and closing of channel gates, activation and inactivation of enzymes, etc. The entire conformational repertoire of a given query protein may not be known; however, it may be possible to infer unknown conformations from other proteins. We developed the ConTemplate method to exploit the richness of the Protein Data Bank (PDB)[ 1 ] for this purpose. ConTemplate uses a three-step process to suggest alternative conformations for a query protein with one known conformation [ 2 ]. First, ConTemplate uses GESAMT to scan the PDB for proteins that share structural similarity with the query [ 3 ]. Next, for each of the collected proteins, additional known conformations are detected using BLAST [ 4 ], and clustered into a predefined number of clusters [ 5 ]. Finally, MODELLER [ 6 ] builds models of the query in various conformations, each representative of a cluster. We demonstrate the application of ConTemplate with S100A6, a member of the S100 family of Ca binding proteins. The vast majority of proteins in this family bind Ca through helix-loop-helix EF-hand motifs. The structure of the protein includes four helices connected by three loops. Calcium binding is coupled to a conformational change, in which helix 3 changes its orientation with respect to helix 4 (Figure 1A and 1B ) [ 7 ]. Helix 2 also changes its positioning with respect to the rest of the protein upon calcium binding, but the change is not as dramatic. The RMSD between the Ca -bound and -free conformations is 4.46Å. The EF-hand motif is found in many PDB entries. Yet, known structures of the Ca -free conformation are relatively rare. These features make the protein an interesting example for examining how the performance of ConTemplate is affected by the distribution of conformations in the PDB: The highly abundant Ca -bound conformation may populate a very large cluster, which could mask the Ca -free conformation. Thus, finding the latter conformation could be challenging. ConTemplate results demonstrated using the S100A6 Ca binding protein . The Ca -free ( A ) and - bound ( B ) conformations are shown in the upper panels; helix 3 is marked in red, and the calcium ions in magenta. C . Reproducing the Ca -bound conformation, starting from the Ca -free conformation as a query. The maximal RMSD between the query and similar proteins is set to 1.2Å, the minimal Q-score to 0.4, and the number of clusters is set to 2. D . Reproducing the Ca -free conformation, starting from the Ca -bound conformation as a query. The similarity cutoffs are the same as in C , the number of clusters is set to 17. Starting from the Ca -free conformation as a query, it is sufficient to set the number of clusters at 2 to retrieve both the Ca -bound and -free conformations. ConTemplate reproduces the Ca -bound conformation with RMSD of 1.6Å (Figure 1C ). This is based on the query's structural similarity to the Ca -free conformation of another member of the family, the S100A2 protein [ 8 ], and the bound conformation of this protein [ 9 ]. The sequence identity between the two proteins is 47%. When the number of clusters is set to be larger than 2, each cluster represents either the Ca -bound or the Ca -free conformation. On the other hand, using the abundant Ca -bound conformation as a query, even with up to three clusters, the process retrieves only variants of the (initial) bound conformation. Only when the number of clusters is four or larger do we obtain at least one cluster representing the Ca -free conformation. In general, the ability to predict the other conformation improves as the number of clusters increases. For example, with 17 clusters, 4 clusters represent the rare conformation, and ConTemplate reproduces the Ca -free conformation with RMSD of 2.43Å (Figure 1D ). This is based on the query's structural similarity to the bound conformation of another member of the family, the S100A12 protein [ 10 ], and the known free conformation of this protein [ 11 ]. The sequence identity between the query and the template is 42%. ConTemplate suggests putative conformations for a query protein with at least one known structure, based on the query's structural similarity to other proteins. In principle, the clustering method enables the detection of distinct conformations, including local conformational changes. However, it may be necessary to adjust ConTemplate's parameters to reveal such changes, especially when looking for rare conformations. When ConTemplate suggests models that are similar to the query, and the clusters are very large, this may indicate that less-common conformations of the query are masked by highly-abundant conformations. Increasing the number of clusters may enable the rarer conformations to be detected. When the additional conformation is not known, it is not trivial to detect the "correct" conformation among the suggested models. A careful examination of the similar proteins and their conformational changes can be useful towards selecting the most probable conformations for the query. In addition, if the number of clusters is large enough, a pathway between the query conformation and a putative conformation may be found, with other models serving as intermediates. Identification of such a pathway could provide insight into the physiological relevance of a newly-detected conformation.

Protein-protein interfaces have been evolutionarily-designed to enable transduction between the interacting proteins. Thus, we hypothesize that analysis of the dynamics of the complex can reveal details about the nature of the interaction, and in particular whether it is obligatory, i.e., persists throughout the entire lifetime of the proteins, or not. Indeed, normal mode analysis, using the Gaussian network model, shows that for the most part obligatory and non-obligatory complexes differ in their decomposition into dynamic domains, i.e., the mobile elements of the protein complex. The dynamic domains of obligatory complexes often mix segments from the interacting chains, and the hinges between them do not overlap with the interface between the chains. In contrast, in non-obligatory complexes the interface often hinges between dynamic domains, held together through few anchor residues on one side of the interface that interact with their counterpart grooves in the other end. In automatic analysis, 117 of 139 obligatory (84.2%) and 203 of 246 non-obligatory (82.5%) complexes are correctly classified by our method: DynaFace. We further use DynaFace to predict obligatory and non-obligatory interactions among a set of 300 putative protein complexes. DynaFace is available at: http://safir.prc.boun.edu.tr/dynaface.

Proteins often alternate between several conformations, e.g., active and inactive states of receptors, open and closed states of channels, etc. However, in many cases only one conformation is known. The prediction of additional (biologically-relevant) conformations of a protein can provide more insight into its function in health and disease. We introduce the ConTemplate computational tool for modeling putative conformations of a query protein with (at least) one known conformation by assuming that pairs of structurally similar proteins may also share similar conformational changes. A three-step procedure is used (Fig. 1 ): First, the protein databank [ 1 ] is searched for structurally similar proteins to the query [ 2 ]. Structure-based pairwise sequence-alignments are built between the query protein and each of the structurally similar proteins. Second, other known conformations (i.e., different from those resembling the query) of these proteins are indicated [ 3 ]. Third, by using the alignments found in the first step, and modeling on the structural templates found in the second, ConTemplate suggests new conformations for the query protein. ConTemplate methodology, demonstrated using the known structure of the EGFR kinase domain in its inactive conformation as a query and reproducing its active conformation; the RMSD between the active and inactive conformations is 4.17Å. Step 1: Selecting proteins with structural similarity to the query; only one is shown here. Step 2: Finding alternative conformations of the proteins detected in Step 1. The black arrows mark the regions with the main differences between the conformations. Step 3: Modeling putative new conformations of the query using the conformations detected in step 2 as templates; only one is shown here. The black arrows indicate the similarities between the model, template and actual known conformation in the main regions of the conformational changes. We demonstrate the method with the kinase domain of the Epidermal Growth Factor Receptor (EGFR). Using the inactive conformation as our query, we reproduce the active conformation [ 4 ] with root mean square deviation (RMSD) of 1.76Å, based on the query's structural similarity to the inactive conformation of Abl tyrosine-kinase [ 5 ], together with the known active conformation of the latter kinase [ 6 ]. The sequence identity between the two kinase domains is only 40%, and the fact that they share similar active and inactive conformations might not be obvious. The idea of inferring new conformations of a protein of interest based on known conformations in related proteins is not new. However, to the best of our knowledge, ConTemplate is the first automated implementation of this approach.

Cyclic nucleotide-gated (CNG) ion channels are nonselective cation channels, essential for visual and olfactory sensory transduction. Although the channels include voltage-sensor domains (VSDs), their conductance is thought to be independent of the membrane potential, and their gating regulated by cytosolic cyclic nucleotide-binding domains. Mutations in these channels result in severe, degenerative retinal diseases, which remain untreatable. The lack of structural information on CNG channels has prevented mechanistic understanding of disease-causing mutations, precluded structure-based drug design, and hampered in silico investigation of the gating mechanism. To address this, we built a 3D model of the cone tetrameric CNG channel, based on homology to two distinct templates with known structures: the transmembrane (TM) domain of a bacterial channel, and the cyclic nucleotide-binding domain of the mouse HCN2 channel. Since the TM-domain template had low sequence-similarity to the TM domains of the CNG channels, and to reconcile conflicts between the two templates, we developed a novel, hybrid approach, combining homology modeling with evolutionary coupling constraints. Next, we used elastic network analysis of the model structure to investigate global motions of the channel and to elucidate its gating mechanism. We found the following: (i) In the main mode of motion, the TM and cytosolic domains counter-rotated around the membrane normal. We related this motion to gating, a proposition that is supported by previous experimental data, and by comparison to the known gating mechanism of the bacterial KirBac channel. (ii) The VSDs could facilitate gating (supplementing the pore gate), explaining their presence in such 'voltage-insensitive' channels. (iii) Our elastic network model analysis of the CNGA3 channel supports a modular model of allosteric gating, according to which protein domains are quasi-independent: they can move independently, but are coupled to each other allosterically.

This special issue hosts the proceedings of the joint 21st annual meeting of Intelligent Systems for Molecular Biology (ISMB) and 12th European Conference on Computational Biology (ECCB), which will take place in Berlin, Germany, between July 21 and 23, 2013 (http://www.iscb.org/ismbeccb2013). This official conference of the International Society for Computational Biology (http://www.iscb.org/), preceded by 11 Special Interest Group (SIG) meetings of 1 or 2 days, two satellite meetings, a Student Council Symposium, Junior Principal Investigators Symposium and two half-day tutorials during July 19–20, 2013, is likely to be the largest annual gathering in the field. The 40 articles in this volume were selected from 233 submissions divided into 13 research areas, led by 26 area chairs. The chairs selected their area’s program committee and oversaw the reviewing process. The chairs included a blend of individuals with experience from previous years and newcomers. Twelve submissions for which area chairs were in conflict were reviewed within ‘Applied Bioinformatics’ headed jointly by the two area chairs and the proceedings chair; four were accepted. Areas, co-chairs and acceptance data are listed in Table 1. Areas, co-chairs and acceptance data Areas, co-chairs and acceptance data A two-tire review procedure was used, which is different from previous years. Of the total submissions, 89 were accepted to round two, including some with conflicting opinions. Of these, 85 were submitted to the second round. The final decision was made based on the revised submissions, taking into consideration the authors response to the reviewer reports, as well as reviewer reports on the revisions. Only 40 articles could be accommodated in the proceedings, forcing us to turn down 35, many of which were of high quality. I was hoping that the authors of most of these would be granted the opportunity to submit revised manuscripts to Bioinformatics, where they could be evaluated while their review history is considered. Unfortunately this opportunity was granted only to 11 manuscripts; I hope that if submitted, these will be readily accepted. More importantly, I would like to extend my apology to the authors of the remaining 24 who went through highly demanding and painful review process. I am hopeful that the authors of these manuscripts found the reviewer comments helpful. The division into areas was somewhat different this year. Two new areas were defined: ‘Metabolic Networks’ and ‘RNA Bioinformatics’ and chairs and reviewers with suitable expertise recruited. ‘Databases and Ontologies’ and ‘Text Mining’ were merged into a single area. All in all 368 members of the bioinformatics community provided reviews. Most articles were assigned to three reviewers and some to four or more. There was significant discussion of the merits of the articles first between referees and the area chairs, and then between area chairs and the proceedings chair. The review also included two conference calls with the area chairs. I am hopeful that the revised review process, which resembles that of journal articles, helped us to correctly choose the most suitable articles for the proceedings, but there is no guarantee. I thank the area chairs and reviewers for their hard work in maintaining a professional review process; it has been harder and longer this year. I thank Steven Leard’s team for extensive technical support with the EasyChair submission and reviewing system; the team at Oxford University Press for type-setting the articles; Conference Chairs Burkhard Rost, Anna Tramontano and Martin Vingron, as well as the ISMB 2013 Steering Committee, for their valuable input; and Steven Leard for helping us oversee the process. Special thanks to Mona Singh for her valuable notes from 2010 and to Bonnie Berger for many good advices at critical moments.

MOTIVATION: Accurate prediction of protein stability is important for understanding the molecular underpinnings of diseases and for the design of new proteins. We introduce a novel approach for the prediction of changes in protein stability that arise from a single-site amino acid substitution; the approach uses available data on mutations occurring in the same position and in other positions. Our algorithm, named Pro-Maya (Protein Mutant stAbilitY Analyzer), combines a collaborative filtering baseline model, Random Forests regression and a diverse set of features. Pro-Maya predicts the stability free energy difference of mutant versus wild type, denoted as ΔΔG. RESULTS: We evaluated our algorithm extensively using cross-validation on two previously utilized datasets of single amino acid mutations and a (third) validation set. The results indicate that using known ΔΔG values of mutations at the query position improves the accuracy of ΔΔG predictions for other mutations in that position. The accuracy of our predictions in such cases significantly surpasses that of similar methods, achieving, e.g. a Pearson's correlation coefficient of 0.79 and a root mean square error of 0.96 on the validation set. Because Pro-Maya uses a diverse set of features, including predictions using two other methods, it also performs slightly better than other methods in the absence of additional experimental data on the query positions. AVAILABILITY: Pro-Maya is freely available via web server at http://bental.tau.ac.il/ProMaya. CONTACT: [email protected]; [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Programs that evaluate the quality of a protein structural model are important both for validating the structure determination procedure and for guiding the model-building process. Such programs are based on properties of native structures that are generally not expected for faulty models. One such property, which is rarely used for automatic structure quality assessment, is the tendency for conserved residues to be located at the structural core and for variable residues to be located at the surface. RESULTS: We present ConQuass, a novel quality assessment program based on the consistency between the model structure and the protein's conservation pattern. We show that it can identify problematic structural models, and that the scores it assigns to the server models in CASP8 correlate with the similarity of the models to the native structure. We also show that when the conservation information is reliable, the method's performance is comparable and complementary to that of the other single-structure quality assessment methods that participated in CASP8 and that do not use additional structural information from homologs. AVAILABILITY: A perl implementation of the method, as well as the various perl and R scripts used for the analysis are available at http://bental.tau.ac.il/ConQuass/. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SUMMARY: The iDBPs server uses the three-dimensional (3D) structure of a query protein to predict whether it binds DNA. First, the algorithm predicts the functional region of the protein based on its evolutionary profile; the assumption is that large clusters of conserved residues are good markers of functional regions. Next, various characteristics of the predicted functional region as well as global features of the protein are calculated, such as the average surface electrostatic potential, the dipole moment and cluster-based amino acid conservation patterns. Finally, a random forests classifier is used to predict whether the query protein is likely to bind DNA and to estimate the prediction confidence. We have trained and tested the classifier on various datasets and shown that it outperformed related methods. On a dataset that reflects the fraction of DNA binding proteins (DBPs) in a proteome, the area under the ROC curve was 0.90. The application of the server to an updated version of the N-Func database, which contains proteins of unknown function with solved 3D-structure, suggested new putative DBPs for experimental studies. AVAILABILITY: http://idbps.tau.ac.il/

Potassium (K+) ion channels switch between open and closed conformations. The nature of this important transition was revealed by comparing the X-ray crystal structures of the MthK channel from Methanobacterium thermoautotrophicum, obtained in its open conformation, and the KcsA channel from Streptomyces lividans, obtained in its closed conformation. We analyzed the dynamic characteristics and energetics of these homotetrameric structures in order to study the role of the intersubunit cooperativity in this transition. For this, elastic models and in silico alanine-scanning mutagenesis were used, respectively. Reassuringly, the calculations manifested motion from the open (closed) towards the closed (open) conformation. The calculations also revealed a network of dynamically and energetically coupled residues. Interestingly, the network suggests coupling between the selectivity filter and the gate, which are located at the two ends of the channel pore. Coupling between these two regions was not observed in calculations that were conducted with the monomer, which emphasizes the importance of the intersubunit interactions within the tetrameric structure for the cooperative gating behavior of the channel.

MOTIVATION: Motion in transmembrane (TM) proteins plays an essential role in a variety of biological phenomena. Thus, developing an automated method for predicting and simulating motion in this class of proteins should result in an increased level of understanding of crucial physiological mechanisms. We have developed an algorithm for predicting and simulating motion in TM proteins of the alpha-helix bundle type. Our method employs probabilistic motion-planning techniques to suggest possible collision-free motion paths. The resulting paths are ranked according to the quality of the van der Waals interactions between the TM helices. Our algorithm considers a wide range of degrees of freedom (dofs) involved in the motion, including external and internal moves. However, in order to handle the vast dimensionality of the problem, we employ some constraints on these dofs in a way that is unlikely to rule out the native motion of the protein. Our algorithm simulates the motion, including all the dofs, and automatically produces a movie that demonstrates it. RESULTS: Overexpression of the RTK ErbB2 was implicated in causing a variety of human cancers. Recently, a molecular mechanism for rotation-coupled activation of the receptor was suggested. We applied our algorithm to investigate the TM domain of this protein, and compared our results with this mechanism. A motion pathway that was similar to the proposed mechanism ranked first, and motions with partial overlap to this pathway followed in rank order. In addition, we conducted a negative-control computational-experiment using Glycophorin A. Our results confirmed the immobility of this TM protein, resulting in degenerate paths comprising native-like conformations.

MOTIVATION: The analysis of co-evolving residues has been exhaustively evaluated for the prediction of intramolecular amino acid contacts in soluble proteins. Although a variety of different methods for the detection of these co-evolving residues have been developed, the fraction of correctly predicted contacts remained insufficient for their reliable application in the construction of structural models. Membrane proteins, which constitute between one-fourth and one-third of all proteins in an organism, were only considered in few individual case studies. RESULTS: We present the first general study of correlated mutations in alpha-helical membrane proteins. Using seven different prediction algorithms, we extracted co-evolving residues for 14 membrane proteins having a solved 3D structure. On average, distances between correlated pairs of residues lying on different transmembrane segments were found to be significantly smaller compared to a random prediction. Covariation of residues was frequently found in direct sequence neighborhood to helix-helix contacts. Based on the results obtained from individual prediction methods, we constructed a consensus prediction for every protein in the dataset that combines obtained correlations from different prediction algorithms and simultaneously removes likely false positives. Using this consensus prediction, 53% of all predicted residue pairs were found within one helix turn of an observed helix-helix contact. Based on the combination of co-evolving residues detected with the four best prediction algorithms, interacting helices could be predicted with a specificity of 83% and sensitivity of 42%. AVAILABILITY: http://webclu.bio.wzw.tum.de/helixcorr/

MOTIVATION: ConSeq is a web server for the identification of biologically important residues in protein sequences. Functionally important residues that take part, e.g. in ligand binding and protein-protein interactions, are often evolutionarily conserved and are most likely to be solvent-accessible, whereas conserved residues within the protein core most probably have an important structural role in maintaining the protein's fold. Thus, estimated evolutionary rates, as well as relative solvent accessibility predictions, are assigned to each amino acid in the sequence; both are subsequently used to indicate residues that have potential structural or functional importance. AVAILABILITY: The ConSeq web server is available at http://conseq.bioinfo.tau.ac.il/ SUPPLEMENTARY INFORMATION: The ConSeq methodology, a description of its performance in a set of five well-documented proteins, a comparison to other methods, and the outcome of its application to a set of 111 proteins of unknown function, are presented at http://conseq.bioinfo.tau.ac.il/ under 'OVERVIEW', 'VALIDATION', 'COMPARISON' and 'PREDICTIONS', respectively.

Abstract Summary: We recently developed algorithmic tools for the identification of functionally important regions in proteins of known three dimensional structure by estimating the degree of conservation of the amino-acid sites among their close sequence homologues. Projecting the conservation grades onto the molecular surface of these proteins reveals patches of highly conserved (or occasionally highly variable) residues that are often of important biological function. We present a new web server, ConSurf, which automates these algorithmic tools. ConSurf may be used for high-throughput characterization of functional regions in proteins. Availability: The ConSurf web server is available at:http://consurf.tau.ac.il. Contact: [email protected]@[email protected]@[email protected]@[email protected] Supplementary Information: A set of examples is available at http://consurf.tau.ac.il under ‘GALLERY’. * To whom correspondence should be addressed.

Abstract Motivation: A number of proteins of known three-dimensional (3D) structure exist, with yet unknown function. In light of the recent progress in structure determination methodology, this number is likely to increase rapidly. A novel method is presented here: ‘Rate4Site’, which maps the rate of evolution among homologous proteins onto the molecular surface of one of the homologues whose 3D-structure is known. Functionally important regions often correspond to surface patches of slowly evolving residues. Results: Rate4Site estimates the rate of evolution of amino acid sites using the maximum likelihood (ML) principle. The ML estimate of the rates considers the topology and branch lengths of the phylogenetic tree, as well as the underlying stochastic process. To demonstrate its potency, we study the Src SH2 domain. Like previously established methods, Rate4Site detected the SH2 peptide-binding groove. Interestingly, it also detected inter-domain interactions between the SH2 domain and the rest of the Src protein that other methods failed to detect. Availability: Rate4Site can be downloaded at: http://ashtoret.tau.ac.il/ It is implemented as a web server at: bioinfo.tau.ac.il/ConSurf Contact: [email protected]@[email protected]@ashtoret.tau.ac.il Supplementary Information: Multiple sequence alignment of homologous SH2 domains, the corresponding phylogenetic tree and additional examples are available at http://ashtoret.tau.ac.il/~rebell Keywords: rate variation among sites; evolutionary conservation; protein evolution; maximum likelihood; SH2 domains. *To whom correspondence should be addressed. †These authors contributed equally.