After evaluation results are extracted, we generate figures describing the results of our experiment. There are a total of six figures (four main and two supplemental). All figure PNGs are found in the figures folder.
- Main figure 1: This figure describes our workflow and displays an image montage of the wildtype and null NF1 genotype single cells, which are hard to distinguish just by eye.
- Main figure 2: This figure shows how subtle the morphological differences are between NF1 genotypes at both the well-population and single-cell levels, which supports are reasoning to pursue a machine learning methodology.
- Main figure 3: This figure shows the results of the model evaluations (precision-recall, accuracy, and confusion matrices) as extracted in the second module of this repository.
- Main figure 4: This figure looks at the feature importances of the model when predicting NF1 genotype. There are two image montages that show six example single-cells (three with the highest values of the feature and three with the lowest), one for each of the top features for predicting each genotype.
- Supplemental figure 1: This figure is an extension of main figure 2, which facets the plot by plate to show that the subtle differences between NF1 genotype are consistent.
- Supplemental figure 2: This figure shows the distributions of FOVs across blur (PowerLogLogSlope) and saturation (PercentMaximal) metrics and where the thresholds were assigned to detect poor-quality images.