Changing an insect’s decision to land requires an array of disciplines. These papers connect machine learning, insect neuroscience, chemical ecology, and the experiments that turn careful observation into behavioral data.
12 papers · 4 fields
Chithrananda, Grand & Ramsundar · 2020 · arXiv preprint
A model can learn from chemical structures before it learns a specific prediction task. ChemBERTa explores this approach on molecular properties. The question for our work is whether that knowledge, combined with behavioral data, can help us make better predictions about compounds we have never tested.
Chepurwar et al. · 2019 · Chemical Senses
Uses similarities between receptor sequences to predict responses to odors, including in flies and mosquitoes. This is a computational baseline for narrowing the search and choosing experiments. Predicting a receptor’s response is one step; understanding what the living animal does requires another.
Tumkaya et al. · 2022 · eLife
In controlled fruit-fly experiments, activating most individual olfactory neuron classes did not produce attraction or avoidance. Responses to pairs were also difficult to predict from their parts. It points us toward interactions between sensory signals, and back to experiments that measure the animal’s behavior.
Task et al. · 2022 · eLife
Maps overlapping chemoreceptor expression within fruit-fly olfactory neurons, with evidence of co-expression in Anopheles mosquitoes. Understanding which receptors share a neuron gives us a better starting point for testing how a compound becomes a neural signal.
Liu, Zhang et al. · 2020 · Current Biology
In diamondback moths, Or35 and Or49 respond to odors from cruciferous plants. Disrupting both receptors removed the tested odor preference for egg laying. The study connects compound, receptor, and behavior in one experimental chain, a model for understanding how an insect chooses a crop.
Pacalon, Audic et al. · 2023 · Nature Communications
Combines molecular simulations, receptor mutations, and electrical recordings to investigate how the synthetic compound VUAA1 enters and binds to Drosophila Orco. It shows how a computational hypothesis about molecular recognition can be put to an experimental test.
Abraham et al. · 2015 · Environmental Entomology
A blend of 11 raspberry volatiles attracted spotted wing drosophila in laboratory assays, but less strongly than raspberry extract. An insect encounters a chemical environment. We need to understand how individual compounds work within the mixtures that help it find a host.
Kirkpatrick et al. · 2018 · Environmental Entomology
Summer females avoided geosmin in laboratory choice tests; winter females showed no significant avoidance. The same compound can produce a different response as an insect’s physiology changes. That makes the animal’s state part of the question we need to test.
Wang et al. · 2021 · Pest Management Science
Pairs behavioral assays with sensory recordings to study plant-derived repellents in two Drosophila species. Responses varied with the compound, dose, and species. It is a reason to test candidate compounds in the insect we need to understand, at the exposure we intend to use.
Conroy et al. · 2024 · Pest Management Science
In enclosed strawberry trials, repellent dispensers reduced fly emergence from fruit close to the dispensers. The effects depended on distance and compound. Bringing a promising chemical beyond the lab means understanding its release rate, its reach, and whether the result holds in a broader field setting.
Steck et al. · 2012 · Scientific Reports
Joins controlled odor delivery with automated tracking of individual walking flies. Concentration, timing, movement, and response latency become part of the same measurement. It offers a way to separate a chemical’s effect from variation in exposure and the animal’s baseline activity.
Geissmann et al. · 2017 · PLOS Biology
Brings together open hardware, video tracking, and software for scalable behavioral experiments. The lesson for our work is to design measurement and analysis together, so that running more experiments produces comparable records of individual behavior.