Yeast Databases to Query
Compiled Knowledge Sources:
MIPS Functional Classification
MIPS Subcellular Localization
MIPS Protein Complexes
MIPS Protein Classes
GO Molecular Function
GO Biological Process
GO Cellular Component
Yeast Two Hybrid - Uetz et al.
Yeast Two Hybrid - Ito et al. (core)
Yeast Two Hybrid - Ito et al. (full)
Synthetic Genetic Array - Tong et al.
MDS Proteomics Complexes - Ho et al.
Cellzome Complexes - Gavin et al.
Proteome Localization - Kumar et al. (observed)
Proteome Localization - Kumar et al. (predicted)
Essentiality and Morphology - Giaever et al.
Yeast Fitness Data - Giaever et al.
FunSpec (an acronym for "Functional Specification") inputs a list of yeast gene names,
and outputs a summary of functional classes, cellular localizations, protein complexes,
etc. that are enriched in the list. The classes and categories evaluated were downloaded
from the MIPS Database and
the GO Database . In addition, many published
datasets have been compiled to evaluate enrichment against. Hypertext links to the publications
The p-values, calculated using the hypergeometric
distribution, represent the probability that the intersection of given list with any
given functional category occurs by chance. The Bonferroni-correction divides
the p-value threshold, that would be deemed significant for an individual test, by the
number of tests conducted and thus accounts for spurious significance due to multiple
testing over the categories of a database. After the Bonferroni correction, only those
categories are displayed for which the chance probability of enrichment is lower than:
p-value/#CD where #CD is the number of categories in the selected database. Without the
Bonferroni Correction, all categories are displayed for which the same probability of
enrichment is lower than: p-value threshold in an individual test
many genes are contained in many categories, especially in the MIPS database (which are
hierarchical) and that this can create biases for which FunSpec currently
makes no compensation. Also the databases are treated as independent from one another,
which is really not the case, and each is searched seperately, which may not be optimal
for statistical calculations. Nonetheless, we find it useful for sifting through the
results of clustering analysis, TAP pulldowns, etc.
For more information, click
Instructions on how to use FUNSPEC can be found