Cell Death

Documentation Index

Fetch the complete documentation index at: https://docs.revilico.bio/llms.txt

Use this file to discover all available pages before exploring further.

​

Why Use This Engine?

In the documentation below, we will use Revilico’s Cell Death assay engine to predict the mode and magnitude of drug-induced cell death across a cancer cell line panel. Rather than measuring bulk viability, this assay decomposes the death response into apoptotic and necrotic fractions and reports caspase 3/7 activity, providing mechanistic insight into how a compound kills cells and whether it engages the programmed apoptotic pathway.

​

Background

Cell death from cytotoxic drugs occurs through two primary mechanisms. Apoptosis is a programmed, energy-dependent process characterized by caspase activation, membrane blebbing, and ordered DNA fragmentation. It is the preferred mechanism for cancer drugs because it is immunologically quiet and does not trigger inflammation. Necrosis is a passive, uncontrolled death caused by membrane rupture and organelle swelling, typically associated with high-dose toxicity or cellular stress exceeding the capacity for organized apoptotic signaling. Distinguishing these modes is clinically relevant: drugs that induce predominantly apoptotic death are generally better tolerated than those that induce necrosis. RevAssay’s Cell Death module wraps the Revilico Virtual Cell GNN+MLP viability model (R-squared 0.87, trained on GDSC1+2 and DepMap CCLE 24Q4) to derive a cellular stress signal, then applies mechanistic scaling to produce apoptosis percentages, necrosis percentages, and caspase 3/7 activity readouts that reflect the dose-dependent induction of each death pathway.

​

Simulation Model

The viability prediction from the GNN+MLP model is converted to a cellular stress signal: $$\text{stress} = 1 - v(c)$$ Where $v(c)$ is the predicted fractional viability at concentration $c$. The apoptotic and necrotic fractions are derived from this stress signal with pathway-specific scaling factors: $$\text{Apoptotic\%} = \text{stress} \times 74 + \varepsilon_a, \quad \varepsilon_a \sim \mathcal{N}(0, 6^2), \quad \text{clamped to } [0, 95]$$ $$\text{Necrotic\%} = \text{stress} \times 14 + \varepsilon_n, \quad \varepsilon_n \sim \mathcal{N}(0, 4^2), \quad \text{clamped to } [0, 25]$$ $$\text{Viable\%} = 100 - \text{Apoptotic\%} - \text{Necrotic\%}$$ The scaling coefficients reflect the biology of EGFR-driven and generic cytotoxic stress: the majority of stress-induced death in epithelial cancer cell lines occurs through the intrinsic apoptotic pathway (caspase 9 then caspase 3/7 activation), with necrosis emerging only at extreme stress levels. Caspase 3/7 bioluminescence is modeled as proportional to the apoptotic fraction: $$\text{Caspase RLU} = \text{Apoptotic\%} \times 11 + \varepsilon_c + 5, \quad \varepsilon_c \sim \mathcal{N}(0, 15^2)$$ The intercept of 5 RLU represents baseline caspase activity in untreated cells.

​

Parameters

​

Outputs

• Apoptotic %: Fraction of cells undergoing programmed apoptotic death at each concentration
• Necrotic %: Fraction of cells undergoing uncontrolled necrotic death
• Viable %: Fraction of surviving cells
• Caspase 3/7 RLU: Luminescence proxy for executioner caspase activation
• Dose-response curves: Per-cell-line curves for each readout across the concentration range
• 96-well heatmap: Plate-view of apoptotic or viable fraction colored by death mode
• Stacked bar charts: Apoptotic (orange), necrotic (red), and viable (green) fractions per cell line per concentration