TABLE 2.
Summarizes the state-of-the-art discussed works together with the employed methods, the cell painting (CP) data used, and the prediction task they were applied to. The superscript numbers in the CP Data column refer to the data set summaries in Table 1. Method abbreviations: Random Forest (RF), Neural Network (NN), Feed-forward NN (FNN), Convolutional NN (CNN), conditional Generative AdversarialNetwork (cGAN).
| Authors | Method | CP Data | Application |
|---|---|---|---|
| Schneidewind et al. (2020) | Similarity search | in house9 | MoA annotation |
| Akbarzadeh et al. (2022) | Similarity search | in house8 | Target annotation |
| Pahl et al. (2023) | Similarity search and clustering | in house3 | Target Annotation |
| Seal et al. (2021) | RF | BBBC0471 | Assay activity |
| Seal et al. (2022) | RF | BBBC0471 | Mitochondrial toxicity |
| Seal et al. (2023a) | RF | BBBC0471 | Cardiotoxicity |
| Moshkov et al. (2023) | Chemprop | BBBC0471 | Assay activity |
| Simm et al. (2018) | RF, k-NN, Macau, FNN | in house | Assay activity |
| Hofmarcher et al. (2019) | FNN, CNN | BBBC0471 | Assay activity |
| Wong et al. (2023) | CNN | in-house5 | Assay activity |
| Tian et al. (2023) | FNN, CNN, classical ML | in-house5 | Assay activity |
| Nguyen et al. (2023) | Contrastive learning | JUMP2 | Molecular property prediction |
| Sanchez-Fernandez et al. (2023) | Contrastive learning | BBBC0471 | Assay activity, cross-modal retrieval |
| Gabriel et al. (2023) | Contrastive learning | JUMP2 | Cross-modal retrieval |
| Zapata et al. (2023) | cGAN | BBBC0471 | Molecular generation |
| Palma et al. (2023) | Autoencoder | BBBC0214 | CP image generation |