bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.64898/2026.04.05.716273

An AI-Assisted Workflow for Reconstruction, Extension, and Calibration of Quantitative Systems Pharmacology Models.

Goryanin, I.·Checkley, S.·Demin, O.
View original source ↗Export metadata ↓

Abstract

AbstractsO_ST_ABSBackgroundC_ST_ABSQuantitative systems pharmacology (QSP) models provide mechanistic insight into drug response but are limited by labor-intensive, expert-driven workflows. We developed an AI-assisted QSP (AI-QSP) framework that integrates large language models (LLMs) with SBML-based modeling to enable automated reconstruction, extension, and calibration of mechanistic models. MethodsThe framework was applied to a published CAR-T QSP model. The model was reconstructed in SBML and extended via LLM-guided prompts to incorporate key resistance mechanisms: T-cell exhaustion, PD-1/PD-L1 checkpoint regulation, and tumor antigen escape. Model development followed an iterative expert-in-the-loop workflow. The resulting model (21 reactions, 9 species) was calibrated to synthetic benchmark data using 19-parameter optimization. Model credibility was assessed using ASME V&V 40 and ICH M15 principles, including global sensitivity and profile-likelihood analyses. ResultsThe calibrated model reproduced benchmark dynamics with high accuracy (mean log-RMSE = 0.132). Sensitivity analysis identified CAR-T killing and bystander cytotoxicity as dominant drivers of tumor response. Profile-likelihood analysis showed 71% of parameters were practically identifiable, with remaining parameters prioritised for future data-driven refinement. ConclusionsAI-assisted QSP modeling enables reproducible, scalable model reconstruction and evolution while maintaining mechanistic transparency and regulatory alignment. This framework provides a foundation for accelerating model-informed drug development in cell and gene therapies.

Explore related subjects

systems biologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Goryanin, I., Checkley, S., Demin, O.. 2026-04-07. An AI-Assisted Workflow for Reconstruction, Extension, and Calibration of Quantitative Systems Pharmacology Models.. https://doi.org/10.64898/2026.04.05.716273

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Peptide-level assay of protein interaction using combinatorial native and denatured proteolysis

Protein protein interactions underlie many extracellular defense and protease regulatory processes, yet identifying their structural interfaces directly in complex biological samples remains challenging. Limited proteolysis coupled with mass spectrometry provides peptide level readouts of local protease accessibility, but linking condition dependent peptide recovery to structural regions in endogenous proteomes requires complementary analysis. Here, we report peptide level assay of protein interaction (PLAPI) using combinatorial native and denatured proteolysis. A single human serum specimen was analyzed in three paired technical preparations using brief trypsin digestion under native conditions and extended digestion after denaturation, followed by isotopic diethyl labeling and LCMS/MS. Applying a 2 of 3 detection rule yielded 920 denatured only peptides and 124 peptides detected under both conditions, with no native only peptides. Mapping to the reference structures placed 13 denatured only C3b peptides at the Factor B binding interface and identified multiple denatured only A2M intersubunit contact regions. Within the A2M bait region, adjacent peptides exhibited different detection patterns; two were detected under both conditions, whereas one was detected only after denaturation. STRING association and enrichment analyses linked the candidate proteins to extracellular defense and protease regulation. These results indicate that PLAPI can provide peptide level evidence for prioritizing structural regions in complex serum samples and guide focused follow up analyses.

systems biology↗
bioRxiv2026

Parsimonious Creatures: Building Gene Regulatory Networks Robust to Somatic Evolution

For early multicellular life to have formed and maintained cellular organization, preventing loss of controlled cell replication due to somatic evolution was of critical importance, and remains important to this day for control of cancer. We study how gene regulatory networks (GRNs) can be designed to slow the divergence of somatic clonal lineages in the context of early multicellular life. We model the mutation of GRNs within simplified 12-celled organisms we term "parsimonious creatures", who must replicate from a single cell to 12 cells and then prevent replication past 12 cells to maximize fitness. We found two minimal motifs can achieve these goals, but both are prone to breakdown through somatic evolution: The feedforward motif, which is loosely regulating but relatively robust, and the simple step motif, which uses strong interactions and feedback to form memoryless ultrasensitivity, but is fragile. These simple motifs may be duplicated to form redundant networks via one of two structures: averaging redundancy, which averages multiple sources of information, and min redundancy, which adopts the most conservative source. Averaging redundancy uniformly slows somatic evolution, while min redundancy stalls somatic evolution until neutral mutations accumulate. Either may be superior dependent on life history. However, averaging redundancy slowly increases fitness as the network becomes larger, but large min redundancy networks eventually become "paralyzed", increasing the probability of senescence and reducing fitness. These GRN designs are idealized, yet provide imperfect solutions with fundamental tradeoffs, offering insight as to why somatic evolution remains an incompletely solved problem for multicellular life.

systems biology↗
bioRxiv2026

Homogeneous network inference for subtype-stratified master regulator analysis in cancer

Master-regulator analysis of 34 TCGA tumour cohorts (cancer x molecular subtype; n between 27 and 562 tumours) requires one regulatory network per cohort. Mutual-information (MI) networks inferred with ARACNe3 are the usual choice, but they are not available for the smallest cohorts; filling the gap with correlation networks would mix two methods inside the very comparison between subtypes that the analysis is meant to make. We therefore infer every network with a single procedure, fixed-size Spearman regulons (FSR): rank correlation between regulator and target, a fixed effect-size threshold rho >= 0.2, the per-target data-processing-inequality (DPI) pruning implemented in corto, bootstrap support, and truncation to the 100 best-supported targets per regulator. We quantify what this choice concedes and what it provides, labelling each claim as a standard result, a heuristic, or an empirical result. Rank correlation is more powerful than the MI estimators we simulated for n < 200 but cannot see non-monotone dependence, which in a synthetic worst case costs at most about 0.05-0.10 of the master regulators recovered; the per-target DPI is a heuristic with a measurable disagreement rate with the MI-based one; and pairwise dependence is not direct interaction, for MI and correlation alike. Where both methods are feasible (n >= 95) they overlap in 0.7 of the top-20 regulators, the same as two filterings of the same ARACNe3 network, although the rank correlation of the full score vector shows a measurable method effect in the tail. Subsampling large real cohorts indicates that networks inferred from fewer than 60 tumours do not reliably reproduce their own cohort's master regulators, that 60-100 is a grey zone, and that a cancer-wide network does not rescue small subtypes. We present FSR as an adequate and uniform choice for multi-cohort analysis, not as a statistically superior one.

systems biology↗