Combinatorial perturbation structure enables graded inference of signaling activity by CIRCE

Cells decode external inputs into intracellular activities of varying magnitude, and inferring both input identity and strength is crucial for cell-fate engineering. However, perturbation-response models typically treat all cells exposed to a given input as a discrete class, ignoring cell-to-cell variation in response magnitude. Here, we introduce a structured-label framework that infers both input identity and strength from combinatorial perturbation screens without dose-resolved training labels. We represent each perturbation combination in a binary combinatorial space and replace one-hot supervision with a Boltzmann distribution that decays with the Hamming distance between combinations. This approach preserves classification accuracy while improving calibration and resolving continuous pathway activities across individual cells. Importantly, intermediate scores primarily reflect ambiguity in cellular response states rather than insufficient model knowledge. Applied to engineered and natural tissues, the calibrated model successfully recovered graded activities. These results establish combinatorial perturbation screens as a self-calibrating framework for decoding graded cellular responses.

Authors

Institutions

Publication Details

Journal
bioRxiv (Cold Spring Harbor Laboratory)
Published
2026-11-20
DOI
https://doi.org/10.64898/2026.11.19.767979
Primary Topic
Gene Regulatory Network Analysis
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Combinatorial perturbation structure enables graded inference of signaling activity by CIRCE

Nicholas T. Hutchins, Pulin Li, Samuel Maffa, Claire Lu
2 links
bioRxiv (Cold Spring Harbor Laboratory)
Gene Regulatory Network Analysis
preprint

Combinatorial perturbation structure enables graded inference of signaling activity by CIRCE

Nicholas T. Hutchins, Pulin Li, Samuel Maffa, Claire Lu
preprint en

Abstract

Cells decode external inputs into intracellular activities of varying magnitude, and inferring both input identity and strength is crucial for cell-fate engineering. However, perturbation-response models typically treat all cells exposed to a given input as a discrete class, ignoring cell-to-cell variation in response magnitude. Here, we introduce a structured-label framework that infers both input identity and strength from combinatorial perturbation screens without dose-resolved training labels. We represent each perturbation combination in a binary combinatorial space and replace one-hot supervision with a Boltzmann distribution that decays with the Hamming distance between combinations. This approach preserves classification accuracy while improving calibration and resolving continuous pathway activities across individual cells. Importantly, intermediate scores primarily reflect ambiguity in cellular response states rather than insufficient model knowledge. Applied to engineered and natural tissues, the calibrated model successfully recovered graded activities. These results establish combinatorial perturbation screens as a self-calibrating framework for decoding graded cellular responses.

bioRxiv (Cold Spring Harbor Laboratory)
Whitehead Institute for Biomedical Research (US), Massachusetts Institute of Technology (US)
National Science Foundation
Gene Regulatory Network Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon