CRAIG is a gene prediction program developed at the Computer and Information Science Department at
the University of Pennsylvania, based on conditional
random fields and online
large-margin training. For the time being only a
human model has been trained but other models will follow soon.
You can submit your own sequences for prediction using our web server. User help and documentation related to
our webserver's use can be accessed here.
You can view the training and development sets used
in developing CRAIG's human gene model. You can also download some CRAIG predictions of human genes.
You can download
here the latest release of craig and lless, a general library for
training and testing linear structure models for sequences (pronounced
"l-less"). Provided along with the full source code is also craig's human
model. Both components are released under GPL v2.0
The research on which this software is based
was funded by the National Science Foundation under ITR grants EIA 0205456
(Language, Learning, and Modeling Biological Sequences) and IIS 0428193 (Machine
Learning for Sequences and Structured Data).