Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Cradle-VAE: Enhancing Single-Cell Gene Perturbation Modeling with Counterfactual Reasoning-based Artifact Disentanglement
Authors: Seungheun Baek, Soyon Park, Yan Ting Chok, Junhyun Lee, Jueon Park, Mogan Gim, Jaewoo Kang
AAAI 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results demonstrate that this approach improves not only treatment effect estimation performance but also generative quality as well. Table 2 shows the quantitative results on the four Perturb-seq datasets. According to the results, CRADLE-VAE overall surpassed all of its baselines in the three evaluation metrics that measure the model s ability to accurately predict cellular responses. |
| Researcher Affiliation | Collaboration | 1Department of Computer Science, Korea University, Seoul, South Korea 2Department of Biomedical Engineering, Hankuk University of Foreign Studies, Yongin, South Korea 3AIGEN Sciences, Seoul 04778, South Korea |
| Pseudocode | Yes | Algorithm 1: CRADLE-VAE Encoding Process. Algorithm 2: CRADLE-VAE Decoding Process. Algorithm 3: CRADLE-VAE Generative Process. |
| Open Source Code | Yes | Code https://github.com/dmis-lab/CRADLE-VAE |
| Open Datasets | Yes | We evaluated CRADLE-VAE on four Perturb-seq datasets, i.e. Norman dataset (Norman et al. 2019), Dixit dataset (Dixit et al. 2016), Replogle dataset (Replogle et al. 2022), and Adamson dataset (Adamson et al. 2016). |
| Dataset Splits | Yes | For datasets involving multi-gene perturbations, the test set was constructed using combinations not encountered during training, representing approximately 25% of the total possible combinations. Conversely, for datasets involving single perturbations, the evaluation emphasized the models ability to capture trends in the observed data within the context of single-perturbation scenarios. |
| Hardware Specification | No | The paper does not provide specific hardware details such as CPU/GPU models or memory specifications used for experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | No | The paper describes experiment settings including datasets, baselines, and evaluation metrics, but it does not specify concrete hyperparameter values (e.g., learning rate, batch size, epochs, optimizer settings) for training. |