Back

CellSwarm: LLM-Driven Cell Agents Recapitulate Tumor Microenvironment Dynamics and Sense Indirect Genetic Perturbations

Meng, X.; Wang, T.; Dong, Z.; Li, X.; Cui, X.; Wang, L.

2026-02-26 systems biology
10.64898/2026.02.25.707926 bioRxiv
Show abstract

Agent-based models of the tumor microenvironment (TME) traditionally rely on hand-coded rules that cannot generalize beyond their programmed logic. Here we present CELLSWARM, a framework that replaces rule-based cell decision-making with large language model (LLM)-driven autonomous agents. Each simulated cell maintains persistent state, 14 signal pathways, and a memory stream, with an LLM serving as its cognitive core. Using structured knowledge bases for cancer-specific context, CELLSWARM recapitulates TNBC microenvironment composition with fidelity comparable to hand-coded rules (Jensen-Shannon divergence 0.144 vs. 0.146; P=0.012 vs. random, Mann-Whitney U test). Beyond matching rule-based performance, LLM-driven agents demonstrate three capabilities absent from rule-based models: cross-cancer generalization by swapping knowledge base entries, treatment response prediction concordant with clinical data (anti-PD-1: 17.6% simulated vs. 21% clinical), and sensing of indirect genetic perturbations that propagate through intermediate signaling cascades (IFN-{gamma} KO: Agent +15.7% vs. Rules +0.3%; P=0.005). CELLSWARM demonstrates that LLM-driven cell agents can recapitulate and extend TME simulation beyond the reach of hand-coded rules.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

1
Cell Systems
167 papers in training set
Top 0.1%
27.9%
2
Nature Medicine
117 papers in training set
Top 0.1%
18.8%
3
Nature Communications
4913 papers in training set
Top 22%
8.5%
50% of probability mass above
4
Proceedings of the National Academy of Sciences
2130 papers in training set
Top 17%
4.0%
5
npj Digital Medicine
97 papers in training set
Top 1%
3.7%
6
npj Systems Biology and Applications
99 papers in training set
Top 0.6%
3.1%
7
PLOS Computational Biology
1633 papers in training set
Top 12%
2.8%
8
Nature Methods
336 papers in training set
Top 3%
2.6%
9
Nature
575 papers in training set
Top 9%
2.1%
10
Nature Machine Intelligence
61 papers in training set
Top 2%
1.9%
11
npj Precision Oncology
48 papers in training set
Top 0.5%
1.7%
12
Cell Reports
1338 papers in training set
Top 24%
1.7%
13
iScience
1063 papers in training set
Top 17%
1.5%
14
Nature Biomedical Engineering
42 papers in training set
Top 1%
1.3%
15
Cancer Cell
38 papers in training set
Top 1%
1.2%
16
Genome Medicine
154 papers in training set
Top 7%
0.8%
17
eLife
5422 papers in training set
Top 55%
0.8%
18
PLOS ONE
4510 papers in training set
Top 67%
0.8%
19
Molecular Systems Biology
142 papers in training set
Top 2%
0.8%
20
Nature Genetics
240 papers in training set
Top 7%
0.8%
21
Cancer Discovery
61 papers in training set
Top 2%
0.8%
22
Cell Reports Medicine
140 papers in training set
Top 8%
0.7%
23
Bioinformatics
1061 papers in training set
Top 10%
0.7%
24
Science Advances
1098 papers in training set
Top 31%
0.7%
25
Nucleic Acids Research
1128 papers in training set
Top 18%
0.7%
26
Cancer Research
116 papers in training set
Top 4%
0.6%
27
Genome Research
409 papers in training set
Top 5%
0.6%
28
Computers in Biology and Medicine
120 papers in training set
Top 6%
0.5%
29
Cellular and Molecular Bioengineering
21 papers in training set
Top 0.5%
0.5%
30
Scientific Reports
3102 papers in training set
Top 80%
0.5%