Functional classification of protein toxins as a basis for bioinformatic screening

Sci Rep. 2017 Oct 24;7(1):13940. doi: 10.1038/s41598-017-13957-1.

Abstract

Proteins are fundamental to life and exhibit a wide diversity of activities, some of which are toxic. Therefore, assessing whether a specific protein is safe for consumption in foods and feeds is critical. Simple BLAST searches may reveal homology to a known toxin, when in fact the protein may pose no real danger. Another challenge to answer this question is the lack of curated databases with a representative set of experimentally validated toxins. Here we have systematically analyzed over 10,000 manually curated toxin sequences using sequence clustering, network analysis, and protein domain classification. We also developed a functional sequence signature method to distinguish toxic from non-toxic proteins. The current database, combined with motif analysis, can be used by researchers and regulators in a hazard screening capacity to assess the potential of a protein to be toxic at early stages of development. Identifying key signatures of toxicity can also aid in redesigning proteins, so as to maintain their desirable functions while reducing the risk of potential health hazards.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Amino Acid Sequence
  • Cluster Analysis
  • Computational Biology*
  • Databases, Protein
  • Gene Order
  • Models, Molecular
  • Protein Domains
  • Proteins / chemistry
  • Proteins / metabolism*
  • Risk
  • Toxins, Biological / chemistry
  • Toxins, Biological / metabolism*

Substances

  • Proteins
  • Toxins, Biological