Leveraging developmental landscapes for model selection in Boolean gene regulatory networks

Area of Science:

• Systems Biology
• Computational Biology
• Developmental Biology

Background:

• Boolean models are widely used for modeling developmental gene regulatory networks (DGRNs) and cellular identity acquisition.
• Reconstructing DGRNs often results in numerous Boolean functions that can explain observed cell fates (attractors).
• Selecting the correct model from these possibilities is a significant challenge.

Purpose of the Study:

• To leverage the developmental landscape and attractor stability for model selection in Boolean DGRNs.
• To introduce a method for reconstructing more accurate and realistic DGRNs.

Main Methods:

• Utilized mean first passage time (MFPT) to quantify attractor stability and cell state transitions.
• Employed stochastic approaches to estimate MFPT for scalability to large networks.
• Developed an iterative greedy algorithm to search for models respecting expected cell state hierarchies.

Main Results:

• Demonstrated strong correlations between different relative stability measures, highlighting MFPT's utility.
• Showcased that a recent Boolean model of Arabidopsis thaliana root development violates expected cell state hierarchies.
• Found that the developed iterative greedy algorithm successfully identifies models meeting expected hierarchy criteria.

Conclusions:

• The proposed methodology provides tools for selecting accurate Boolean functions in DGRN reconstruction.
• The approach enables the creation of more biologically realistic and precise Boolean models of DGRNs.
• Relative attractor stability, particularly MFPT, is a valuable metric for DGRN model selection and lineage construction.