ESPript/ENDscript: extracting and renderingsequence and 3D information from atomicstructures of proteinsPatrice Gouet * , Xavier Robert and Emmanuel Courcelle 1Laboratoire de BioCristallographie, IBCP-CNRS UMR 5086 UCBL, 7 passage du Vercors, 69367 Lyon Cedex 07,France and1 Laboratoire de Biologie Mole´culaire et des Relations Plantes Microorganismes, BP 27 Chemin de BordeRouge, 31326 Castanet Tolosan, FranceReceived February 15, 2003; Revised and Accepted March 31, 2003ABSTRACTThe fortran program ESPript was created in 1993, todisplay on a PostScript f i gure multiple sequencealignments adorned with secondary structureelements. A web server was made available in 1999and ESPript has been linked to three major webtools: ProDom which identif i es protein domains,PredictProtein which predicts secondary structureelements and NPS@ which runs sequence alignmentprograms. A web server named ENDscript wascreated in 2002 to facilitate the generation of ESPriptf i gures containing a large amount of information.ENDscript uses programs such as BLAST, Clustaland PHYLODENDRON to work on protein sequencesand such as DSSP, CNS and MOLSCRIPT to workon protein coordinates. It enables the creation,from a single Protein Data Bank identif i er, of amultiple sequence alignment f i gure adorned withsecondary structure elements of each sequence ofknown 3D structure. Similar 3D structures are super-imposed in turn with the program PROFIT and a f i nalf i gure is drawn with BOBSCRIPT, which showssequence and structure conservation along the Catrace of the query. ESPript and ENDscript areavailable at http://genopole.toulouse.inra.fr/ESPript.INTRODUCTIONProteins with sequence identity >30% normally belong to thesame family and have similar conformation and function (1,2).Such clear homologues are likely to have diverged from acommon ancestor and their sequences may show conserveddifferences between species of organisms. Convergent evolu-tion can also occur within a family, if certain homologues havedeveloped additional functions such as the capability to bindligands away from the active site. The side function is alsolikely to be written in the sequence and better still in the 3Dstructure which is generally even more conserved. Thus, thesimultaneous comparison of sequence and structure informa-tion is of importance to detect biological specif i cities in agroup of proteins.The program ESPript, Easy Sequencing in PostScript,generates f i gures of aligned sequences with secondarystructure information (3). It can serve as a tool for structure/function analyses. ESPript reads text outputs from multiplesequence alignment programs such as Clustal (4) andMULTALIN (5), as well as from programs able to identifysecondary structure elements from structure f i les such as DSSP(6) and STRIDE (7). Residues are boxed according to theirsimilarity score and secondary structure elements are drawn atthe top of sequences blocks. ESPript can be used forpublication purposes and the user has access to numerousfeatures to optimize its f i gure (selection of displayedsequences, choice of colours, symbols to highlight chosenresidues). The program is written in Fortran and can beexecuted locally on Linux or Unix machines or on a web servervia a CGI interface.A server named ENDscript was later created to routinelyproduce ESPript f i gures with a maximum of sequence/structure information (8). The query is the code of a proteinstructure deposited with the Protein Data Bank (9) or a f i lewith atomic coordinates. NMR and crystallographic structuresin PDB format are supported. The sequence is extracted fromthe query and ENDscript performs a BLAST search (10)against a selected protein sequence database to detect clearhomologues (search with E-value cutoff of 10 ?6 by default).The result is piped to Clustal or MULTALIN for amultiple sequence alignment. The f i nal ESPript f i gure showsthe aligned sequences with the secondary structure elements ofeach sequence of known 3D structure. Additional informationis presented, such as intermolecular and protein–ligandcontacts detected by CNS (11), accessibility calculated byDSSP and hydropathy. A BOBSCRIPT (12) f i gure is alsogenerated, showing a ribbon representation of the querycoloured according to sequence homology. The molecule canbe rotated via a VRML f i le produced by MOLSCRIPT (13).*To whom correspondence should be addressed. Tel: þ33 472722624; Fax: þ33 472722616; Email: p.gouet@ibcp.fr3320–3323 Nucleic Acids Research, 2003, Vol. 31, No. 13DOI: 10.1093/nar/gkg556Published by Oxford University Press 2003
