RPBS (Ressource Parisienne en Bioinformatique Structurale) is a resource dedicated primarily to structural bioinformatics. It is the result of a joint effort by several teams to set up an interface that offers original and powerful methods in the field. As an illustration, we focus here on three such methods uniquely available at RPBS: AUTOMAT for sequence databank scanning, YAKUSA for structure databank scanning and WLOOP for homology loop modelling. The RPBS server can be accessed at and the specific services at .

SPROUTS (Structural Prediction for pRotein fOlding UTility System) is a new database that provides access to various structural data sets and integrated functionalities not yet available to the community. The originality of the SPROUTS database is the ability to gain access to a variety of structural analyses at one place and with a strong interaction between them. SPROUTS currently combines data pertaining to 429 structures that capture representative folds and results related to the prediction of critical residues expected to belong to the folding nucleus: the MIR (Most Interacting Residues), the description of the structures in terms of modular fragments: the TEF (Tightened End Fragments), and the calculation at each position of the free energy change gradient upon mutation by one of the 19 amino acids. All database results can be displayed and downloaded in textual files and Excel spreadsheets and visualized on the protein structure. SPROUTS is a unique resource to access as well as visualize state-of-the-art characteristics of protein folding and analyse the effect of point mutations on protein structure. It is available at http://bioinformatics.eas.asu.edu/sprouts.html.

Proteins come in all shapes and sizes. Although it is possible to predict with reasonable success their structure from their sequence, the process of folding a chain of amino acids into its tertiary structure remains partially understood. This article addresses several characteristics pertaining to protein folding. The development of the Most Interacting Residues (MIR) algorithm, which dynamically simulates the early folding events, permits a reasonable ab initio prediction of the deeply buried critical residues involved in the formation of the protein core. The analysis of MIR positions with respect to protein 3D topology, in particular, to fragments called Tightened End Fragments (TEF) that might be good candidate for autonomous folding units, suggests that they are also essential for defining core stability. To validate this hypothesis, this study measures the sensitivity of MIR residues to point mutations. It is performed on a set of 385 proteins from a database that contains stability data calculated with five different algorithms. Tools have been developed to help the analysis and a consensus of the five methods is proposed. It results that positions predicted both as a MIR and a minimum of stability for the consensus are good candidates for the folding nucleus, and consequently their mutations may be hazardous.

Abstract. The development of a method that accurately predicts protein folding nucleus is critical at least on two points. On one hand, they can participate to misfolded proteins and therefore they are related to several amyloid diseases. On the other hand, as they constitute structural anchors, their prediction from the sequence can be valuable to improve database screening algorithms. The concept of Most Interacting Residues (MIR) aims at predicting the amino acids more likely to initiate protein folding. An alternative approach describes a protein 3D structure as a series of Tightened End Fragments (TEF). Their spatially close ends have been shown to be mainly located in the folding nucleus. While the current sequence-driven approach seems to capture all MIR, the structure-driven method partially fails to predict known folding. We present a stability-based analysis of protein folding to increase the recall and precision of these two methods.Results: Prediction of the folding nucleus by MIR algorithm is in agreement with mutation stability prediction.

The MIR algorithm provides an ab initio prediction of a protein's core residues. An improved version, the MIR2, is presented and validated on 3203 proteins from PDB. Structures are decomposed in Closed Loops, their limits constituting the observed core residues. They are predicted by MIR2 with an accuracy approaching 80%.

Database scanning programs such as BLAST and FASTA are used nowadays by most biologists for the post-genomic processing of DNA or protein sequence information (in particular to retrieve the structure/function of uncharacterized proteins). Unfortunately, their results can be polluted by identical alignments (called redundancies) coming from the same protein or DNA sequences present in different entries of the database. This makes the efficient use of the listed alignments difficult. Pretreatment of databases has been proposed to suppress strictly identical entries. However, there still remain many identical alignments since redundancies may occur locally for entries corresponding to various fragments of the same sequence or for entries corresponding to very homologous sequences but differing at the level of a few residues such as ortholog proteins. In the present work, we show that redundant alignments can be indeed numerous even when working with a pretreated non-redundant data bank, going as high as 60% of the output results according to the query and the bank. Therefore the accuracy and the efficiency of the post-genomic work will be greatly increased if these redundancies are removed. To solve this up to now unaddressed problem, we have developed an algorithm that allows for the efficient and safe suppression of all the redundancies with no loss of information. This algorithm is based on various filtering steps that we describe here in the context of the Automat similarity search program, and such an algorithm should also be added to the other similarity search programs (BLAST, FASTA, etc...).