Malaria-MOI: A flexible and scalable tool for predicting multiplicity of infection in malaria parasites

Estimating the multiplicity of infection (MOI) is critical for understanding malaria transmission dynamics and within-host Plasmodium parasite diversity. We developed Malaria-MOI, a fast, flexible, and species-agnostic Python tool that infers MOI directly from standard genomics files (e.g., BAM, VCF format) derived from whole genome sequencing (WGS), without requiring prior training data, curated panels or complex parameter tuning. When benchmarked on 27 Plasmodium falciparum mixed-clone samples with known MOI, Malaria-MOI matched or outperformed leading methods, achieving a root mean squared error of 0.38, mean absolute error of 0.15, and a correlation of 0.78 using genomic variants across 165 diverse loci. These results exceeded the median performance of existing Bayesian and likelihood-based approaches. Applied to 8,208 P. falciparum field samples, Malaria-MOI showed a strong negative correlation with FWS (ρ = − 0.76), consistent with its accurate capture of within-host diversity. It demonstrated high sensitivity for detecting monoclonal infections (0.99) and superior specificity (0.76) compared to estMOI (0.58), particularly at lower sequencing coverage. The tool also identified regional MOI differences aligned with transmission intensity, detecting more polyclonal infections in high-transmission areas such as West Africa, where estMOI underestimated complexity. Overall, Malaria-MOI accommodates diverse input types, including whole-genome data and diversity loci, and is compatible with both Illumina and Nanopore sequencing platforms. It is integrated into the Malaria-Profiler framework and is well-suited for genomic surveillance of Plasmodium and other pathogens, especially when elucidating transmission intensity for disease elimination activities. https://github.com/LSHTMPathogenSeqLab/Malaria-MOI.