Developing an Agent-Based Drug Model to Investigate the Synergistic Effects of Drug Combinations

Abstract

The growth and survival of cancer cells are greatly related to their surrounding microenvironment. To understand the regulation under the impact of anti-cancer drugs and their synergistic effects, we have developed a multiscale agent-based model that can investigate the synergistic effects of drug combinations with three innovations. First, it explores the synergistic effects of drug combinations in a huge dose combinational space at the cell line level. Second, it can simulate the interaction between cells and their microenvironment. Third, it employs both local and global optimization algorithms to train the key parameters and validate the predictive power of the model by using experimental data. The research results indicate that our multicellular system can not only describe the interactions between the microenvironment and cells in detail, but also predict the synergistic effects of drug combinations.

Development of an Early Prediction Model for Subarachnoid Hemorrhage With Genetic and Signaling Pathway Analysis

“…For example, the SAH intervention experiment sample size was too small for us to demonstrate high predictive accuracy for the model. In future work, we will integrate more recent bioinformatics research algorithms (Zhang et al, 2016(Zhang et al, , 2017a(Zhang et al, , 2018(Zhang et al, , 2019aGao et al, 2017; and data into the system to overcome the problems.…”

“…Second, we use E-Bayes (Carlin and Louis, 2010), SVM-RFE (Duan et al, 2005), SPCA (Zou et al, 2006), and statistical tests (Zhang et al, 2016(Zhang et al, , 2018(Zhang et al, , 2019b(Zhang et al, ,d, 2020Xiao et al, 2019) to investigate key genes from experimental data by considering both SAH and LCN2 as factors. Third, we integrate the logistic regression (LR), support vector machine (SVM), and Naive-Bayes algorithms (Xia et al, 2017;Zhang et al, 2017aZhang et al, , 2019a into an ensemble learning model (Gao et al, 2017;Zhang et al, 2019b) to build a model for early SAH prediction.…”

Leveraging Mathematical Modeling to Quantify Pharmacokinetic and Pharmacodynamic Pathways: Equivalent Dose Metric

“…These methods have great utility in discovering and quantifying drug interactions; however, they cannot be leveraged to understand the mechanisms underlying the identified synergy/antagonism. While other methods have leveraged mechanistic data to identify synergy (Al-Lazikani et al, 2012; Gao et al, 2017; Yin et al, 2018), the proposed equivalent dose framework provides quantitative mechanistic insight into intracellular drug effects and allows for predictions of treatment response under a variety of treatment conditions. We posit that this mechanistic approach could facilitate clinical translation of combination therapies.…”

Developing an Embedding, Koopman and Autoencoder Technologies-Based Multi-Omics Time Series Predictive Model (EKATP) for Systems Biology research

“…After we employ 20 different proteomics time series datasets to test the KAE and EKATP, Table 1 shows the predictive error of the KAE and EKATP at 1,000 steps under different initial noise and complexity ( ) conditions, which demonstrates that the EKATP has less of a statistically significant minimum, maximum, average and variance of the predictive error than the KAE under each noise and complexity ( ) condition ( p -value <0.05) ( Gao et al, 2017 ; Li et al, 2017 ; Gao et al, 2021 ). Table 1 implies that the EKATP has statistically significant predictive power for different time series datasets.…”

“…However, it is inaccurate to predict the future state of genomics time series with nonlinear complicated interactions because the Lorentz system is not good at processing nonlinear complicated interactions ( Lai et al, 2018 ). Currently, delay embedding theory ( Sauer et al, 1991 ; Holmes et al, 2012 ) is commonly used to transform the spatial information (complicated interactions) into temporal information (the future state of the time series ( Chen et al, 2020 )) for dimensional reduction ( Gao et al, 2017 ; Li et al, 2017 ; Xia et al, 2017 ; Zhang et al, 2019b ; Zhang et al, 2019c ; Wu et al, 2020 ; You et al, 2020 ; Zhang et al, 2021b ), whereas Koopman theory ( Koopman, 1931 ) can switch the nonlinear system into a linear system to reduce computing cost. Therefore, our first research question asks if we can develop such a time series predictive model that integrates the Lorentz system with delay embedding and Koopman theory to accurately predict the future state of genomics time series with chaotic behavior.…”