< Back to Article

Comparative Analysis of Yeast Metabolic Network Models Highlights Progress, Opportunities for Metabolic Reconstruction

Fig 4

Evaluating model predictions of single-gene essentiality.

Flux balance analysis was conducted to predict whether individual genes were essential for growth using seven different media formulations and two different model biomass objective functions for each model. Gene essentiality predictive performance is summarized in this table by the Matthews’ Correlation Coefficient (MCC). Model predictions were compared to two reference lists of essential genes: one derived from the saccharomyces genome database (SGD-based gene list) and one from Kuepfer et al. (Kuepfer-based gene list). These lists are provided as Supplementary Information. Modeled medium formulations included each model’s default medium, (Medium: Default), a minimal glucose-limited medium (Medium: Min-Glu), a synthetic complete glucose-limited medium based on Kennedy et al. (Medium: SC-Glu), and a synthetic medium based on Kuepfer et al. using glucose (Medium: Kuepfer-Glu), galactose (Medium: Kuepfer-Gal), glycerol (Medium: Kuepfer-Gly), or ethanol (Medium: Kuepfer-Eth) as carbon source. Simulations were conducted using each model’s default biomass definition (Biomass: Default) or the iFF708 model biomass definition (Biomass: iFF). In this heat map, color intensity is based upon positive Matthews’ Correlation Coefficient (MCC) (no parameter combinations lead to negative MCCs for any model), each row is a unique set of model parameters, and models are arranged in chronological order from left to right.

Fig 4