Hist-i-fy—a multiple histidine post-translational-modification (PTM) prediction server based on protein sequences using convolution neural network: a case study on mass spectroscopy data

Abstract

Computational characterization of multiple Histidine (His) post-translational-modifications (PTM) at enzyme active sites complements tedious experimental characterization in proteins-of-unknown-functions (PUFs) and domain-of-unknown-functions (DUFs). There are only a handful of Histidine-PTM-prediction-tools and those also annotate only a single function. Here, we addressed the problem using artificial neural networks on functional histidine dataset curated from enzyme (protein) sequences available in UniProt database (sample size n = 1584). The convolution-neural-network (CNN) model (‘Hist-i-fy’) performed the best with 75% overall accuracy/F1-score. A case study was performed on histidine-phosphorylation (n = 34) obtained from mass spectroscopy data. For the first time, we report multiple His-PTM-prediction-tool (https://histify.streamlit.app/& https://github.com/dibyansu24-maker/Histify), with optimal performance. The inputs to the tool are (i) protein sequence containing histidine, and (ii) the histidine residue number. Prediction output is one out of the eight histidine functions—acetylation, ribosylation, glycosylation, hydroxylation, methylation, oxidation, phosphorylation, and protein splicing.

Communicated by Ramaswamy H. Sarma

Keywords:

• Histidine post-translational-modifications (PTM)
• Convoluted Neural Network (CNN)
• protein sequence
• UNIPROT database

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.