Web supplement to
"The Chemical Genomic Portrait of Yeast: Uncovering a Phenotype for All Genes"

Maureen E. Hillenmeyer, Eula Fung, Jan Wildenhain, Sarah E. Pierce, Shawn Hoon, William Lee, Michael Proctor, Bob P. St. Onge, Mike Tyers, Daphne Koller, Russ B. Altman, Ronald W. Davis, Corey Nislow, and Guri Giaever.

Data download


Please find more detailed descriptions for these files in the Supplementary materials.

Analyzed fitness defects


The following files contain the fitness defect scores. Each row represents a gene deletion strain, given by its ORF name and deletion batch. Each column represents a treatment experiment. Explanation of experiment names can be found on Page 4 of the Supplementary Materials.

Clustering genes


In Figure 3 of the manuscript, we displayed three clusters extracted from genome-wide clustering. We performed this genome-wide clustering by:

Multi-drug resistance (MDR) genes
and sensitivity counts for all deletion strains


For each deletion strain, we counted the number of unique conditions in which it was sensitive. We defined a gene as a "multi-drug resistance" gene if the deletion strain was sensitive in >20% of unique small molecule conditions (excluding environmental stress conditions).


Raw Affymetrix .CEL data


The following files contain the raw .CEL data from the Affymetrix TAG3 microarrays. Each experiment has an associated set of controls to which it was compared. The mapping from each experiment to its associated controls are denoted in the key files.

Scripts for analyzing .CEL data


Download scripts and associated files
Three perl scripts are used to parse the CEL data, perform normalization and assign fitness defect scores.
  1. The first script, "raw_file_data.pl" maps the raw intensity data in each input CEL file to their associated strain-tags.
  2. A second script, "normalize_data.pl" takes the output produced by the first script and normalizes it so that the data for each experiment has a mean inte nsity of 1500.
  3. The last script, "fitness_profile.pl", calculates both fitness defect log ratios and significance values (1 or 0) for a given set of control and treatmen t experiments using the normalized data produced by the second script.
Run 'perldoc script_name' to see more information about each script.




Inquiries can be addressed to guri.giaever@utoronto.ca.