• Downloads


Welcome to the chemical genomics website. This website provides an interface for querying the chemical genomics data. Valid queries include compound names (Benomyl), gene names (TUB3) or ORFs (YML124C), and Gene Ontology terms (tubulin complex assembly) or IDs (GO:0007021). It is possible to query multiple compounds, genes, or GO terms by separating them with a semicolon (;).

As an example query, try out: "Benomyl; TUB3; Tubulin Complex Assembly;"


Citation: Piotrowski, J. S., Li, S. C., Deshpande, R., Simpkins, S. W., Nelson, J., Yashiroda, Y., . . . Osada, H., Yoshida, M., Myers, C. L., Boone, C. (2017). Functional Annotation of Chemical Libraries across Diverse Biological Processes. Nature Chemical Biology. Advance online publication. doi:10.1101/112557

Nelson, J., Simpkins, S. W., Safizadeh, H., Li S. C., Piotrowski, J. S., . . . Osada, H., Yoshida, M., Boone, C., Myers, C. L. MOSAIC: a chemical-genetic interaction data repository and web resource for exploring chemical modes of action. bioRxiv. doi:10.1101/112854


Database Update: July 27, 2017
Website Update: July 28, 2017

  • Conditions
  • Genes
  • GO Terms

Select a condition from the following table.

Picture Name Alt. Name Conc. Formula SMILES Bioactivity Molecular Weight Dataset Collection Links

Select a gene from the following table.

Standard Name Systematic Name Gene Name Gene Alias Links

Select a gene ontology term from the following table.

GO Name GO Description
  • Process-level Target Predictions
  • Gene-level Target Predictions
  • Chemical Genetic Profile
  • ChemSIM
  • ProfSIM
Process-level Target Prediction

Process-level target predictions are the Gene Ontology terms which are enriched among genes which have a similar genetic interaction profile to the chemical genetic interaction profile. For example, Benomyl has an enrichment for tubulin complex assembly because the chemical genetic interaction profile for Benomyl is similar to the genetic interaction profiles for GIM3, GIM4, GIM5, YKE2, CIN1, CIN2, and PAC10, which are all annotated to tubulin complex assembly.

Empirical p-values and z-scores are calculated by calculating a mean gene target prediction score for genes within a GO term and comparing it to a null distribution. Two false discovery rates were calculated based on comparing the predicted z-score to control compound chemical genomic profiles or randomized resampled chemical genomic profiles.

Predictions in white are considered high confidence (FDR<25). Predictions in gray are outside of the high confidence threshold.

Chemical Name GO Name GO ID p-value z-score FDR_DMSO FDR_RAND
Gene-level Target Prediction

Gene-level target predictions are the genes which have similar genetic interaction profiles to the compound of interest. Gene-level target prediction dotcosine scores are calculated by computing a cosine score between the chemical genetic interaction profile and a normalized genetic interaction profile. For example, the chemical genetic interaction profile of Benomyl is similar to its supposed protein target's genetic interaction profile, TUB3.

Compounds in white have at least one high confidence (FDR<25) process-level target prediction. Compounds in gray do not have any high confidence predictions.

Chemical Name Gene Common Name Gene ORF Name dotcosine
Chemical Genetic Profile

These are the chemical genetic interaction sensitivities (-) and resistances (+). The chemical genetic interaction scores are a z-score which represents the standardized deviation of each strain in the condition profile with respect to its counterpart strain in the reference DMSO profiles.

Compounds in white have at least one high confidence (FDR<25) process-level target prediction. Compounds in gray do not have any high confidence predictions.

Chemical Name Gene Common Name Gene ORF Name Score
Chemical Similarity

The chemical similarity or structural similarity is calculated using the All-Shortest Path (ASP) structure descriptor (depth 8) along with the Braun-Blanquest similarity coefficient.

Compounds in white means chemical 2 have at least one high confidence (FDR<25) process-level target prediction. Compounds in gray means chemical 2 does not have any high confidence predictions.

Chemical 1 Chemical 2 Score
Profile Similarity

Profile similarity or functional similarity is measured using a pearson correlation on the chemical genetic interaction profiles.

Compounds in white means chemical 2 have at least one high confidence (FDR<25) process-level target prediction. Compounds in gray means chemical 2 does not have any high confidence predictions.

Chemical 1 Chemical 2 Score