Whole Organism High-Content Screening by Label-Free, Image-Based Bayesian Classification for Parasitic Diseases
(A) Phenotype scores produced by the Bayesian prediction model from 5 replicate plates containing controls (M169/DMSO) or anti-schistosome drugs (10 µg/ml), praziquantel (PZQ), oltipraz (OPZ), methylclonazepam (MCZ), Ro15-5458 (Ro15), oxamniquine (OX) and dihydroartemisinin (DHA). N = 120 wells per condition and significance values were measured by non-parametric one-way ANOVA with individual drug treatment being compared to the DMSO controls wells by a Dunn's comparison post test, *** = p<0.001. NS; not significant. (B) Z factor analysis based on the distribution of the negative (DMSO; circle) and positive controls (OPZ; triangle) from 15 analyzed plates of anti-schistosome compounds (dotted lines represent the upper and lower thresholds). (C) Typical larval phenotypes and phenotype scores induced by the anti-schistosome drug treatments or media (M169) and solvent (DMSO) controls. (D) Drug treatment class prediction for wells treated with the anti-schistosome drugs and the mean number of correctly predicted larvae per well.