Tags
AAM, Biomarker, Bond Changes, Classification, Directed evolution, EBI, EC Number, EMBL-EBI, Enzyme Classification, Functional genomics, Genes, Github, GO, KEGG, Metabolic pathways, Metabolomics, MMP, Novel Genes, Nutrigenomics, Ontology, Predicting, Protein Sequence, Rational design, Reaction Similarity, Seq2EC, Sequence, Small Molecule, substructure, Toxicity
Next Generation Sequencing (NGS) data is knocking at our door and simultaneously, our ability to design novel enzymes (rational design or directed evolution) using high throughput methods has improved tremendously. As a result, the demand to link enzymatic sequences to their chemical products and metabolic pathways is ever increasing. On the other hand, the push to generate Metabolomics data to design Biomarkers, understand Toxicity, Functional genomics and Nutrigenomics has given researchers a run for their money!
Last year we launched EC-Blast (my old post), a robust tool to compare chemical reactions using chemical knowledge of bond changes, molecule molecule pair (MMP) and molecule substructures. This tool helps plough through and understand the reactions present in the Enzyme Commission (E.C.) classification. This has generated a lot of interest in the research community and industry to revisit and mine the knowledge which might have been overlooked by traditional methods. Feedbacks from our users strongly suggested a demand for tools/methods to systematically link the protein sequences to the knowledge of bond changes, molecule molecule pair (MMP) and molecules substructures.
We have recently developed Sequence to Enzyme (Seq2EC) a novel tool (Figure 1) to:
- Detect novel enzymatic genes- test if a sequence has potential enzymatic function.
- Predict GO terms, E.C. number, bond changes and MMP from a protein sequence.
- Predict potential genes and protein domains from chemical bonds (e.g. O-C, O-P formed or cleaved).
Figure 1: Predicting enzyme function from protein sequence
There are two ways to use the tool:
- Seq2EC webservice/webserver (requires patience): http://www.ebi.ac.uk/thornton-srv/software/rblec/
- Download and run locally (fast) – contact me!
The tool reports a star ranking schema (Table 1) for predicted enzymes which is based on the statistical confidence of finding domains, bonds, catalytic sites etc.
Table 1: Star ranking schema explained
| Star | Rank | Note |
| **** | Platinum | Confident till the 4th level of EC |
| *** | Gold | Confident till the 3rd level of EC |
| ** | Silver | Confident till the 2nd level of EC |
| * | Bonze | Confident till the 1st level of EC |
| . | Ambiguous | Might have an enzymatic activity |
Please go ahead and test your favourite sequence(s) or chemical changes and send me your feedback/comments!
Asad's Blog 
Good job! Impressive! Since you accept UniProt identifiers as input for prediction, have you considered making a pre-calculated list of predictions for these available from your website?
Thank you!
One could pre calculate all the entries in the UniProt but given the size of the genomic data (virus, bacteria and eukaryotes etc), I have left it to the users discretion to run it on a case by case basis….on the other hand the output will be indeed novel and useful 🙂
I agree! However, depending on the time it takes for calculating these, the UniProt resource could well benefit from this extra annotation. Would they perhaps be willing to set up a link to your predictor?