Data-driven inference of digital twins for high-throughput phenotyping of motile and light-responsive microorganisms

Abstract

Light can affect the movement of microorganisms. These responses can drive collective behaviours like photoaccumulation and photodispersion, which play a key role in broader biological functions like photosynthesis. Our understanding of these emergent phenomena is severely limited by difficulties in obtaining data needed to establish accurate models that can serve as a basis for multi-scale analyses. Here, we address this issue by developing an integrated experimental and computational platform to collect large temporal imaging datasets that allow for the inference of digital twins -- mathematically precise computational models that accurately mirror the behavior of individual microorganisms -- and show that they can replicate the light response of diverse microorganisms in silico. We show that a generalised phenomenological model of movement can be effectively parameterised from experimental data to capture key behavioural traits of two commonly studied photo-responsive microorganisms (Euglena gracilis and Volvox aureus) and demonstrate our models ability to accurate reproduce patterns of movement for individuals and populations in response to dynamic and spatially varying light patterns. This work takes steps towards the automated phenotyping of multi-scale behaviours in biology and complements high-throughput genome sequencing efforts by allowing for more comprehensive and quantitative genotype-to-phenotype mappings. It also unlocks new opportunities for the design of spatial control algorithms to guide collective microorganism behaviour.