Arjmandi, H. | bioRxiv Science

bioRxiv2023

DeepGEEP: Data-Driven Prediction of Bacterial Biofilm Gene Expression Profiles

Predicting the gene expression profile (GEEP) of bacterial biofilms in response to spatial, temporal, and concentration profiles of stimulus molecules holds significant potential across microbiology, biotechnology, and synthetic biology domains. However, the resource and time-intensive nature of experiments within Petri dishes presents significant challenges. Data-driven methods offer a promising avenue to replace or reduce such experiments given sufficient data. Through wellcrafted data generation techniques, the data scarcity issue can be effectively addressed. In this paper, an innovative methodology is presented for generating GEEP data over a Petri dish that results from a specific chemical stimulus release profile. A twodimensional convolutional neural network (2D-CNN) architecture is subsequently introduced to leverage the synthesized dataset to predict GEEP variations across bacterial biofilms within the Petri dish. The approach, coined DeepGEEP, is applied to data generated by a particle-based simulator (PBS) to enable a flexible evaluation of its efficacy. The proposed method attains a significant level of accuracy in comparison to established benchmark models such as Linear SVM, Radial Basis Function SVM, Decision Tree, and Random Forest.

bioinformatics↗

bioRxiv2023

3D Cell Cultures Amplify Diffusion Signals

Biophysical models can predict the behavior of cell cultures including 3D cell aggregates (3DCAs), thereby reducing the need for costly and time-consuming experiments. Specifically, mass transfer models enable studying the transport of nutrients, oxygen, signaling molecules, and drugs in 3DCA. These models require the defining of boundary conditions (BC) between the 3DCA and surrounding medium. However, accurately modeling the BC that relates the inner and outer boundary concentrations at the border between the 3DCA and the medium remains a challenge that this paper addresses using both theoretical and experimental methods. The provided biophysical analysis indicates that the concentration of molecules inside boundary is higher than that at the outer boundary, revealing an amplification factor that is confirmed by a particle-based simulator (PBS). Due to the amplification factor, the PBS confirms that when a 3DCA with a low concentration of target molecules is introduced to a culture medium with a higher concentration, the molecule concentration in the medium rapidly decreases. The theoretical model and PBS simulations were used to design a pilot experiment with liver spheroids as the 3DCA and glucose as the target molecule. Experimental results agree with the proposed theory and derived properties. Author summaryThe primary objective of our research was to enable the development of reliable biophysical models for three-dimensional cell aggregates (3DCAs). To achieve this goal, we employed a combination of theoretical and experimental methods to derive and characterize the amplification boundary condition (BC), which represents the relation of inner and outer boundary concentrations at the border between a 3DCA and its surrounding medium. By understanding the amplificaiton BC, we can better comprehend the transport and diffusion processes that occur within 3DCAs. The significance of our research lies in its potential to advance the understanding of 3DCAs and their underlying biophysical processes. This knowledge is crucial for a wide range of applications, including drug design and analysis of drug dosages within tissues. This factor may provide insight into the mechanisms behind tumor development and morphogenesis. In particular, the packed structure of cancer tumors enables them to receive and trap a higher concentration of nutrients and oxygen molecules based on the amplification factor. Thus, this study could contribute to the development of novel approaches to manage and treat cancerous tissues.

biophysics↗