FAQ

• License of PEER

PEER by Oliver Stegle, Leopold Parts, Matias Piipari is licensed under the Gnu Public License 2.0. PEER can be freely used by other people as long as the source of PEER is adequately cited.

• Download and install Please see our dedicated install and download section.

• Are the standalone tool, Python and R different implementations?

No, they are different interfaces to the same C++ library. Thanks to SWIG, we can interface to the different alternatives.

• Where can I learn more about the algorithms ?

Underlying PEER are two alternative applications of Bayesian factor analysis-like models. The details are described in these article

• A Bayesian framework to account for complex non-genetic factors in gene expression levels greatly increases power in eQTL studies. (PLoS Computational Biology, May 2010)

• Joint Genetic Analysis of Gene Expression Data with Inferred Cellular Phenotypes. (PLoS Genetics, January 2011)

• How does PEER compare to other methods?

We compared PEER to a number of alternatives in the primary publications. PCA, SVA and ICE are widely used alternative to recover hidden determinants. Unique to PEER is the combination with prior knowledge and a consistent approach to handle covariates and their overlap with the hidden factors. See the two papers above for more details on comparisons with alternative methods.

• Can PEER handle large datasets?

The largest test to date was performed by Brent Pedersen from University of Colorado, who ran PEER on 2.1 million features from 194 individuals. PEER took about 27 hours to finish, and used 10 GB memory.