General information
The interaction dataset used in the paper was derived from two sources: i) raw interaction data from an SGA screen specifically designed for metabolic genes (present study), ii) raw interaction data from a recent large-scale screen employing the same experimental procedure as the present study, but representing genes in all major functional categories, including metabolism (Costanzo et al. 2010). The metabolism-specific SGA screen was constructed by crossing 613 query mutants, including 78 hypomorphic (DAmP) alleles of essential genes, against an array of 470 null mutants. The combined raw dataset for metabolic genes (metabolic screen + Costanzo et al. 2010) contains approximately 417,000 double mutants after quality filtering, including double mutants that have been independently constructed more than once. The combined processed interaction dataset was derived from the raw dataset by applying the following procedures: i) identify double mutants showing significant interaction by employing cutoffs of |ε| > 0.08 and p-value < 0.05, ii) for gene pairs screened more than once: if replicate screens show opposite interaction signs and at least one of them is significant, both pairs were removed; if they show the same interaction sign (both positive or both negative), the interaction with the lowest p-value was retained and both pairs are reported with that interaction. Additionally, we defined a smaller high-confidence dataset in which all gene pairs were independently screened at least twice to minimize false interactions. Here, two genes are considered as interacting if at least one screen shows |ε| > 0.08 and p-value < 0.05, and another screen shows p-value < 0.05 of the same interaction sign, whereas non-interacting pairs are defined as those not showing |ε| > 0.08 and p-value < 0.05 in any of the screens. Any other gene pairs were removed from the high-confidence set.


Supplementary Data File S1. Raw data file from metabolic SGA screen

sgadata_metabolic_array.zip - 7 MB
The file contains the raw data of the SGA screen specifically designed for metabolism in a tab-delimited format with 13 columns:

• Query ORF
• Query gene name
• Array ORF
• Array gene name
• Genetic interaction score (ε)
• Standard deviation
• p-value
• Query single mutant fitness (SMF)
• Query SMF standard deviation
• Array SMF
• Array SMF standard deviation
• Double mutant fitness
• Double mutant fitness standard deviation

Supplementary Data File S2. List of null mutant and hypomorphic alleles used in the combined dataset

all_alleles_included_metabolic_array+Costanzo2010.txt - 7 KB
The file lists all gene deletant and hypomorphic alleles that have been screened as array or query strains in either the metabolic or the Costanzo et al. 2010 large-scale screen and are alleles of metabolic network genes.  Temperature-sensitive (TS) and DAmP queries are indicated by "_tsq" and "_damp" suffixes, respectively.


Supplementary Data File S3. Combined processed interaction dataset

combined_data_ic.zip - 2.5 MB
The file contains the processed interaction dataset that combines data on metabolic network genes from both the metabolism-specific screen and from the Costanzo et al. 2010 large-scale screen. Tab-delimited text file with 5 columns:

• ORF 1
• ORF 2
• Genetic interaction score (ε)
• p-value
• Inferred genetic interaction (negative, positive, none)

Supplementary Data File S4. Combined processed high-confidence interaction dataset

combined_data_highconfidence.zip - 1.5 MB
The file contains the high-confidence processed interaction dataset that combines data on metabolic network genes from both the metabolism-specific screen and from the Costanzo et al. 2010 large-scale screen. This dataset is restricted to gene pairs that have been independently screened more than once and show concordant interactions across replicates (see General Information). Tab-delimited text file with 5 columns:

• ORF 1
• ORF 2
• Genetic interaction score (ε)
• p-value
• Inferred genetic interaction (negative, positive, none)

Supplementary Data File S5. Modified genome-scale metabolic model of yeast

iMM904_NADcorrected.xml - 2 MB
The file contains a genome-scale metabolic model in SBML format that is based on the iMM904 model, but incorporates modifications in the NAD biosynthesis pathway as suggested in the paper (see Supplementary Figure 4).