For questions or comments, please contact Roy Kishony

Additional Links:

Daniel Segre'

Nature Genetics

Flux balance model details

• The S. cerevisiae stoichiometric reconstruction used for this work is from J. Förster, I. Famili, P. Fu, B. Ř. Palsson and J. Nielsen, "Genome-Scale Reconstruction of the Saccharomyces cerevisiae Metabolic Network", Genome Research 2003 13: 244-253:

  • List of reactions

  • List of metabolite names

  • Biomass composition

• Additional information about the yeast stoichiometric reconstruction used in the current work is available here

Single and double knockout yeast FBA data

FILE NAME	CONDITIONS
fitness_data_Nominal.txt	Nominal
fitness_data_O_M.txt	O-
fitness_data_O_P.txt	O+
fitness_data_C_M.txt	C-
fitness_data_C_P.txt	C+
fitness_data_C_PP.txt	C++
fitness_data_AC.txt	AC
fitness_data_NH3.txt	NH3
fitness_data_aa_P.txt	aa+
fitness_data_dS1.txt	dS1
fitness_data_dS2.txt	dS2
fitness_data_cyt.txt	cyt
fitness_data_thr.txt	thr

The conditions are defined in supplementary Fig. 4 of the manuscript.

Each file contains two lists:
1) calculated fitness of all the dingle genes knockouts
2) calculated fitness of all the double gene knockouts.
Note that this list includes non essential single genes (i.e., genes with zero fitness in the list number 1 are excluded in list number 2).

For the data presented in the manuscript, gene products which belong to a common structural cluster where represented by a single gene. Gene cludters were defined according to the definitions of the Mips database

Effectively, this means excluding the genes listed in the file redundant_gene_complexes.txt.

The Prism algorithm

Download prism.m

Supplementary Material

Supplementary Fig. 1
The epistasis scale captures the strength of the epistasis effect relative to natural extreme values.

Supplementary Fig. 2
Analysis of the effects of double deletions of metabolic enzyme genes in simple metabolic networks demonstrating examples of multiplicative, aggravating and buffering gene deletion interactions.

Supplementary Fig. 3
Sensitivity analysis with respect to free parameters and physiological conditions.

Supplementary Fig. 4
Sub-classification of buffering interactions into directional and non-directional links is related to the underlying biochemical network.

Supplementary Fig. 5
Changes in monochromatically interacting epistasis modules following variation of oxygen uptake rate.

Supplementary Fig. 6
Examples of monochromatic clustering of three toy networks using the Prism algorithm.

Supplementary Fig. 7
Randomization algorithms and statistical tests for monochromaticity in the epistasis network.

Supplementary Table 1
List of free parameters.

Supplementary Methods
The yeast flux balance model.

Supplementary Note
Discussion of Prism modules and predicted interactions.

Supplementary References

Supplementary Video 1 (avi 2M)
Schematic demonstration of monochromatic classification. A network of two types of connections, such as buffering (green) and aggravating (red) epistasis, is sorted into module of genes that interact with one another in a purely monochromatic way.