| 1 | = A DDBJ-KEGG-PDBj workflow: from pathways to protein-protein interactions = |
| 2 | |
| 3 | * For the details of the project, see [wiki:DDBJ-KEGG-PDBj]. |
| 4 | |
| 5 | == Members == |
| 6 | * Yasumasa Shigemoto (WABI/SABI; DDBJ, Japan) |
| 7 | * Akira Kinjo (PDBj, Japan) |
| 8 | * Soichi Ogishima (Tokyo Medical and Dental Univ, Japan) |
| 9 | * Masumi Itoh (Hokkaido Univ) |
| 10 | |
| 11 | = The objective = |
| 12 | Given a KEGG pathway ID of a specific species, |
| 13 | 1. Identify homologous proteins of the pathway. |
| 14 | 2. Infer possible protein-protein interactions between components of the homologous pathways. |
| 15 | |
| 16 | = How? = |
| 17 | |
| 18 | 1. Input: KEGG pathway ID |
| 19 | 1. Get amino acid sequences of all the proteins involved in the given pathway. (KEGG API) |
| 20 | 1. For each protein sequence, run BLAST against the !UniProt (!SwissProt / !TrEMBL) database to identify its homologs. (WABI) |
| 21 | 1. Sort the BLAST results according to species. (a local program) |
| 22 | * At this point, we will have a phylogenetic profile. |
| 23 | 1. For each BLAST hit (a !UniProt entry), run BLAST against PDB. (PDBj !SeqNavi API). |
| 24 | 1. Find ''physical'' interactions between different components. (a local program) |
| 25 | * That is, if two (!UniProt) sequences are homologous to different chains of the same PDB entry, they are predicted to be in physical contact. |
| 26 | 1. Output the phylogenetic profile and the predicted species-wise PPI. |
| 27 | |
| 28 | = An example = |
| 29 | |
| 30 | 1. KEGG pathway: path:eco00020 |
| 31 | 1. Phylogenetic profile: |
| 32 | |