EnzML: multi-label prediction of enzyme classes using InterPro signatures

Abstract

BackgroundManual annotation of enzymatic functions cannot keep up with automatic genome sequencing. In this work we explore the capacity of InterPro sequence signatures to automatically predict enzymatic function.ResultsWe present EnzML, a multi-label classification method that can efficiently account also for proteins with multiple enzymatic functions: 50,000 in UniProt. EnzML was evaluated using a standard set of 300,747 proteins for which the manually curated Swiss-Prot and KEGG databases have agreeing Enzyme Commission (EC) annotations. EnzML achieved more than 98% subset accuracy (exact match of all correct Enzyme Commission classes of a protein) for the entire dataset and between 87 and 97% subset accuracy in reannotating eight entire proteomes: human, mouse, rat, mouse-ear cress, fruit fly, the S. pombe yeast, the E. coli bacterium and the M. jannaschii archaebacterium. To understand the role played by the dataset size, we compared the cross-evaluation results of smaller datasets, either constructed at random or from specific taxonomic domains such as archaea, bacteria, fungi, invertebrates, plants and vertebrates. The results were confirmed even when the redundancy in the dataset was reduced using UniRef100, UniRef90 or UniRef50 clusters.ConclusionsInterPro signatures are a compact and powerful attribute space for the prediction of enzymatic function. This representation makes multi-label machine learning feasible in reasonable time (30 minutes to train on 300,747 instances with 10,852 attributes and 2,201 class values) using the Mulan Binary Relevance Nearest Neighbours algorithm implementation (BR-kNN).

Modeling ecological drivers in marine viral communities using comparative metagenomics and network analyses

“…Exclusive reads per the method above for LineP summer photic viromes, summer aphotic viromes, winter photic viromes, and winter aphotic viromes were compared against the similarity matrix of proteins (SIMAP) released on August 20, 2013 (57) using BLASTX (58) to assign function as previously described (22). Briefly, these analyses were implemented using a custom data analysis pipeline written in Perl and bash shell and executed on a high-performance computer using PBSPro (blastpipeline_ simap.tar).…”

“…Interpro ids in the SIMAP functional annotation were mapped to EC numbers using the swissprot_kegg_proteins_ec.csv as a mapping (59) (ipr_to_ec.pl). Read hit counts were normalized based on sequencing effort in the included viromes and converted into ipath2 format (create_ipath.pl) for visual representation in the ipath2 viewer (58).…”

Is EC class predictable from reaction mechanism?

“…Sequenced-based protein function prediction has hundreds of thousands of annotated sequences available [18]. …”

Enzyme function and its evolution

“…Indeed, the circularity of the combined process of propagating annotations and then predicting function, based on the same annotations and homologies, may be problematic. Sequence-based enzyme function predictions based on EC number annotations in databases can indeed give very impressive results [35] and such predictive exercises can be extended to include mechanism [36], both processes usually operating mostly via the detection of homology -although 3D structure-based methods also exist [37,38,39]. Using mechanisms and catalytic chains as defined in MACiE, the corresponding UniProt sequences are interrogated against InterPro signatures [29] to re-express the MACiE entries in terms of the signatures present in them.…”