A paper on ESPERR was published in Genome Research on 19 October 2006, and is available online

ESPERR 7 species RP Scores

These scores are currently available for the two most recent builds of human and of mouse.

They are avaailable in the UCSC genome browser as "Reg Potential 7 species" in the "Expression and Regulation" track group

Raw scores are in UCSC wiggle track format, and are available both as full range log odds scores (*.scores.bz2) and truncated so that all negative scores are set to zero (*.scores.truncated.bz2).


Code is available via anonymous svn. It can be checked out by:

svn co http://coltrane.bx.psu.edu/svn/esperr/trunk/ esperr-trunk

Turnkey template

A template Makefile that can be easily modified to run the ESPERR procedure on arbitrary training data is available in svn under the training_template directory. It requires that the software listed below be built and installed in some location.


The python code is in the directory

and can be built with the standard tools (see the README).

This code requires:

  • Python 2.4
  • Numeric python (currently this must be Numeric, everything will transition to the better maintained numpy in the near future)
  • Scipy (used only for matrix math in the inferrence of ancestral distributions)
  • Pyrex (needed to create C code and wrappers for the various models)
  • Pypar (optional, needed to use the MPI version)
  • The bx-python libraries (see below)


The bx-python libraries used by this software are availabe here: