EBI is an Outstation of the European Molecular Biology Laboratory. Annotation Procedures for Structural Data Deposited in the PDBe at EBI

Established in 1996 at the European Bioinformatics Institute – autonomous structural database capability in Europe. One of the four sites around the world where structural data can be deposited. Stable and clean repository for macromolecular structure data. Services that allow users to access, search and retrieve structural data from a single web access point. The Protein Data Bank in Europe (PDBe) group

Depositor AutoDep4.0 “Raw” PDB file Automated + Manual Curation “Annotated” PDB file Depositor’s comments Structure release Data Processing at PDBe

Data Deposition at the PDBe using AutoDep4.0 Structure deposition and archival tool developed at the PDBe (EBI). Based on Java/XML technology. Available freely under license for academic and industry users. Easy to install and use for in-house archiving before deposition to the PDB via the PDBe interface.

The Curation Process Raw information obtained from the Depositor - a) atomic coordinates (proteins, nucleic acids, Ligands, solvents) b) source of the macromolecule c) number of protein chains present in the asymmetric unit d) experimental data (structure factor file) Three Phases of Curation – 1)Automated Curation 2)Manual Curation 3)Final Checks.

Automated Curation Consists of series of programs written in Fortran and Perl Annotators contribute ideas and programs in order to improve the curation process We work in a Unix command line interface This is the first Step : a big wrapper

The Wrapper Automatically generates: Chain ID for every HETATM and HOH (gets the chain ID of the closest polypeptide chain) Quaternary structure, according to PISA (REM300&350) Structure validation: Close contacts (REM500) and chirality checks Solvent molecules that lie farther than expected from the protein (REM525) HELIX, SHEET, SSBOND, CISPEP records Residue by residue Mapping against the Uniprot database Dohlc output

Contents of a Curated PDB file Sequence related information: 1)Sequences (SEQRES) – all macromolecules present during crystallization, including expression tags and residues missing from the coordinates due to disorder. 2) Sequence Database reference (DBREF) - provides mapping (FASTA alignment) between the sequence (SEQRES) against the Uniprot database.

Macromolecular Structure Database Checks made … Is the Uniprot accession number correct? The sequence similarity between the Uniprot sequence and the target sequence should be minimum ~95% Identification of N- and C-termini cross references with the Uniprot and addition of fragment information (if any) to the COMPND record. Merge the data from the Uniprot entry to COMPND (Molecule name), SOURCE (Scientific name of the organism) and KEYWDS Addition of EC number, if available

Curation procedures continued…. 1) If no Sequence database reference available: the sequence is self- referenced (i.e. the database reference will be the PDB entry itself). 2) Additional details regarding the sequence (gaps, cloning artifact, structural disorder is provided in REMARK 999 3) Disagreement between a Uniprot sequence and the sequence present in the PDB file (SEQADV): marked as a) Engineered Mutation, b) conflict or c) microheterogeneity. 4) Residues missing from the coordinates – listed in REMARK 465 5) Non-hydrogen atoms missing from the coordinates- listed in REMARK 470 6) Zero-occupancy residues - REMARK 475 7) Zero-occupancy atoms - REMARK 480 8) Related PDB entries (same Uniprot Accession numbers) are listed in REMARK 900 9) Backbone discrepancies

Ligand Curation Ligands interacting with a protein/DNA chain → substrate, product, inhibitor (drug molecule), metal ion, modified amino acid or nucleotide. MODRES token added for Modified amino acids and nucleotides which are part of the polymer (i.e. protein/DNA) chain. Specialized software (Do Het Link and Connect records) used to get the bond type, stereochemistry and IUPAC compliant name for each ligand in the structure. DOHLC is a graph based structure comparison algorithm – checks each ligand/HET with dictionary definition, renames residues and atoms. Generates REMARK 620(metal coordination), LINK and CONECT records. DOHLC failing – bad geometry, incomplete ligand or new HETGROUP If no match found for a HETGROUP – new ligand created HETGROUP with missing atoms - REMARK 610 HETGROUP with zero-occupancy atoms – REMARK 615

Generating Assembly Information ASU Contents Expand Crystal Symmetry Analyze surface and contacts Best !! Possible Assemblies Loss of accessible surface area >10% of total surface. True complexes also look good ! Biological unit – Biologically relevant form of the molecule Quaternary structures – the way protein chains tend to associate with one another The matrices forming the quaternary structure are reported as BIOMT records in REMARK 350

1E94PISA assembly PISA assemblies

Structure validation

Macromolecular Structure Database Final Checks: Programs check for PDB format accuracy and internal consistency Manual check by another Annotator Automatic generation of the letter to depositor + Manual addition of special comments