DBgDel: Database-Enhanced Gene Deletion Framework for Growth-Coupled Production in Genome-Scale Metabolic Models

TL;DR

DBgDel introduces a database-enhanced framework to accelerate growth-coupled gene deletion design for genome-scale metabolic networks by mining prior deletion strategies from MetNetComp. It uses a two-step approach: STEP 1 derives a reduced gene pool $G_{ ext{remain}}$ from known maximal deletion strategies, and STEP 2 applies an extended MILP-based gDel_minRN with this pool to quickly identify feasible deletions that satisfy $v_{ ext{growth}} ge GR_{ ext{threshold}}$ and $v_{ ext{target}} ge PR_{ ext{threshold}}$, while maximizing repressed reactions. The method achieves substantial speedups (average 6.1x) and competitive success across models of increasing scale (e_coli_core, iMM904, iML1515), outperforming or matching established approaches in many cases. Analyses show that Predicted-$G_{ ext{remain}}$ often yields the best overall trade-off between speed and success, and extending DBgDel to other algorithms demonstrates flexible applicability. Overall, DBgDel provides a practical, scalable route to rapid design of growth-coupled production strains by leveraging prior knowledge embedded in a gene deletion database.

Abstract

When simulating metabolite productions with genome-scale constraint-based metabolic models, gene deletion strategies are necessary to achieve growth-coupled production, which means cell growth and target metabolite production occur simultaneously. Since obtaining gene deletion strategies for large genome-scale models suffers from significant computational time, it is necessary to develop methods to mitigate this computational burden. In this study, we introduce a novel framework for computing gene deletion strategies. The proposed framework first mines related databases to extract prior information about gene deletions for growth-coupled production. It then integrates the extracted information with downstream algorithms to narrow down the algorithmic search space, resulting in highly efficient calculations on genome-scale models. Computational experiment results demonstrated that our framework can compute stoichiometrically feasible gene deletion strategies for numerous target metabolites, showcasing a noteworthy improvement in computational efficiency. Specifically, our framework achieves an average 6.1-fold acceleration in computational speed compared to existing methods while maintaining a respectable success rate. The source code of DBgDel with examples are available on https://github.com/MetNetComp/DBgDel.

Figures (3)

• Figure 1: A toy example of the constraint-based model where circles and rectangles represent metabolites and reactions, respectively. Black rectangles denote external and internal reactions. $r_1$, $r_2$ correspond to two substrate uptake reactions. $r_8$, $r_{9}$ correspond to cell growth, and target metabolite production reactions, respectively. The reaction rates are constrained by the range $[l_i, u_i]$. This example shows only part of the model, the rest after $m_7$ is omitted.
• Figure 2: An overview of the proposed DBgDel framework. The DBgDel framework comprises two steps: (1) STEP 1, DBgDel takes known gene deletion strategies from the MetNetComp database as input and constructs a remaining gene set $G_{\text{remain}}$ as output; and (2) STEP 2, DBgDel uses an extended version of gDel_minRN algorithm that incorporates the $G_{\text{remain}}$ to calculate the deleted gene set $D$ for a new target metabolite, as the final output gene deletion strategies of the framework.
• Figure 3: The toy model with a new target metabolite. Circles and rectangles represent metabolites and reactions, respectively. Black rectangles denote external and internal reactions. $r_1$, $r_2$ correspond to two substrate uptake reactions. $r_8$, $r_{10}$ correspond to cell growth, and target metabolite production reactions, respectively. The reaction rates are constrained by the range $[l_i, u_i]$.