New study uses mathematical analysis of walking patterns for early detection of Parkinson’s disorder.

The art of crafting insect-specific pesticides using machine learning

Read time: 3 mins
New Delhi
4 Jun 2019
The art of crafting insect-specific pesticides using machine learning

The use of insecticides—chemicals that kill insects destroying crop—has revolutionized agriculture. About 10,000 different species of insects are known to cause havoc to food and fibre crops worldwide, and using insecticides has significantly reduced crop losses. In fact, in 2014 alone, insecticide use accounted for 7.5% of the total pesticide use. However, in the guise of protecting our crops, these chemicals harm the environment, kill agriculturally essential insects like bees, and are hazardous to human health. Can we design an insect-specific insecticide that is effective and does not wipe out biodiversity?

A new study by researchers at the Indraprastha Institute of Information Technology, CSIR-Institute of Microbial Technology and University of Delhi has answers. The researchers have developed a computer-based tool called NeuroPIpred that can help make novel neuropeptides—protein-like molecules used by neurons to communicate—specific to certain insects. The study is published in the journal Scientific Reports and was funded by the Department of Science and Technology.

Neuropeptides are made of about 5-80 amino acids and control a variety of biological activities like growth, development, metabolism, homeostasis as well as behavioural activities like mating, migration, and egg laying. Any disruption in the amino acid sequence of a neuropeptide can cause debilitating effects on the insect, leading to its death. Thus, creating such variant neuropeptides can become effective in killing insect pests and information from NeuroPIpred can help to design insect-specific insecticides with mutant neuropeptides.

NeuroPIpred is a web-based tool that uses machine learning techniques. It takes in the amino acid composition of insect neuropeptides as input, which is provided by a web-based database application called DINeR (Database for Insect Neuropeptide Research). Based on this input, it learns new possible sequences and predicts whether a given peptide is an insect neuropeptide or not, and accurately differentiates between insect neuropeptides and human neuropeptides.

“The tool allows users to discover possible neuropeptide regions in a protein and to design the best mutant of a neuropeptide of desired activity, and hence designing an insect-pest specific neuropeptide antagonist which would be an effective insecticide for insect pest management,'' says Dr. Singh (DU) & Prof. Raghva (IIITD), corresponding authors of the study.  

Conventional insecticides are non-specific, highly toxic and kill most insects, including beneficial insect pollinators like bees, along with the target pests. Since NeuroPIpred can describe the structure and physicochemical properties, like the amino acid residues, of the insect neuropeptide, this information can help design the most apt mutant insect neuropeptide that kills only the target insects. Besides, as neuropeptides are organic chemicals, they do not leave behind any toxic residues, making them environmentally friendly.

The field of designing insecticidal neuropeptides is still nascent and requires further research. "Large-scale identification of neuropeptides across most damaging insect pests, like those belonging to the orders Lepidoptera (butterflies and moths), Coleoptera (beetles), Hemiptera (true bugs) and Diptera (flies) still needs to be done," remark Dr. Raghva & Dr Singh.

“The designing of neuropeptide-based insecticides, with possible enhancement of is characteristics based on its applications, and the formulation of prototypes followed by several field trials could pave the way forward,” they say. 


This article has been run past the researchers, whose work is covered, to ensure accuracy.