Conserved Peptides Recognition by Ensemble of Neural Networks for Mining Protein Data – LPMO Case Study

Mining protein data is a recent promising area of modern bioinformatics. In this work, we suggested a novel approach for mining protein data – conserved peptides recognition by ensemble of neural networks (CPRENN). This approach was applied for mining lytic polysaccharide monooxygenases (LPMOs) in 19 ascomycete, 18 basidiomycete, and 18 bacterial proteomes. LPMOs are recently discovered enzymes and their mining is of high relevance for biotechnology of lignocellulosic materials. CPRENN was compared with two conventional bioinformatic methods for mining protein data – profile hidden Markov models (HMMs) search (HMMER program) and peptide pattern recognition (PPR program combined with Hotpep application). The maximum number of hypothetical LPMO amino acid sequences was discovered by HMMER. Profile HMMs search proved to be more sensitive method for mining LPMOs than conserved peptides recognition. Totally, CPRENN found 76 %, 67 %, and 65 % of hypothetical ascomycete, basidiomycete, and bacterial LPMOs discovered by HMMER, respectively. For AA9, AA10, and AA11 families which contain the major part of all LPMOs in the carbohydrate-active enzymes database (CAZy), CPRENN and PPR + Hotpep found 69–98 % and 62–95 % of amino acid sequences discovered by HMMER, respectively. In contrast with PPR + Hotpep, CPRENN possessed perfect precision and provided more complete mining of basidiomycete and bacterial LPMOs.

Peptide Pattern Recognition for high-throughput protein sequence analysis and clustering

Large collections of protein sequences with divergent sequences are tedious to analyze for understanding their phylogenetic or structure-function relation. Peptide Pattern Recognition is an algorithm that was developed to facilitate this task but the previous version does only allow a limited number of sequences as input.I implemented Peptide Pattern Recognition as a multithread software designed to handle large numbers of sequences and perform analysis in a reasonable time frame. Benchmarking showed that the new implementation of Peptide Pattern Recognition is twenty times faster than the previous implementation on a small protein collection with 673 MAP kinase sequences. In addition, the new implementation could analyze a large protein collection with 48,570 Glycosyl Transferase family 20 sequences without reaching its upper limit on a desktop computer.Peptide Pattern Recognition is a useful software for providing comprehensive groups of related sequences from large protein sequence collections.