Related Experiment Video
Updated: Nov 23, 2025
Constructing and Visualizing Models using Mime-based Machine-learning Framework
Published on: July 22, 2025
CellBox: Interpretable Machine Learning for Perturbation Biology with Application to the Design of Cancer Combination
Bo Yuan1, Ciyue Shen1, Augustin Luna1
1Department of Cell Biology, Harvard Medical School, Boston, MA, USA; cBio Center, Department of Data Sciences, Dana-Farber Cancer Institute, Boston, MA, USA; Broad Institute, Cambridge, MA, USA.
Abstract:
Systematic perturbation of cells followed by comprehensive measurements of molecular and phenotypic responses provides informative data resources for constructing computational models of cell biology. Models that generalize well beyond training data can be used to identify combinatorial perturbations of potential therapeutic interest. Major challenges for machine learning on large biological datasets are to find global optima in a complex multidimensional space and mechanistically interpret the solutions. To address these challenges, we introduce a hybrid approach that combines explicit mathematical models of cell dynamics with a machine-learning framework, implemented in TensorFlow. We tested the modeling framework on a perturbation-response dataset of a melanoma cell line after drug treatments. The models can be efficiently trained to describe cellular behavior accurately. Even though completely data driven and independent of prior knowledge, the resulting de novo network models recapitulate some known interactions. The approach is readily applicable to various kinetic models of cell biology. A record of this paper's Transparent Peer Review process is included in the Supplemental Information.
Insights
This study introduces a hybrid computational approach combining mathematical models and machine learning to analyze cell behavior. The method accurately models cellular responses to drug perturbations, revealing known biological interactions.
Area of Science:
- Computational biology
- Systems biology
- Machine learning in biology
Background:
- Computational models are crucial for understanding complex cell biology.
- Machine learning faces challenges in optimizing and interpreting large biological datasets.
- Developing predictive models for cellular responses to perturbations is key for therapeutic discovery.
Purpose of the Study:
- To develop a hybrid computational framework integrating mathematical models with machine learning.
- To accurately model cellular dynamics and responses to perturbations.
- To address challenges in global optimization and mechanistic interpretation for biological data.
Main Methods:
- A hybrid approach combining explicit mathematical models of cell dynamics with a machine-learning framework (TensorFlow).
- Application of the framework to a melanoma cell line perturbation-response dataset.
- Efficient training of models to describe cellular behavior.
Main Results:
- The hybrid model accurately describes cellular behavior following drug treatments.
- De novo network models generated by the framework recapitulate known biological interactions.
- The approach demonstrates efficient training and accurate modeling of cellular responses.
Conclusions:
- The hybrid approach offers an effective method for constructing and interpreting computational models of cell biology.
- This framework can identify potential combinatorial perturbations for therapeutic development.
- The approach is versatile and applicable to various kinetic models in cell biology.
More Related Videos
07:15Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
09:53Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography
Published on: August 16, 2020
