S-Peaker command-line tool installation:
To generate feature values and run the logistic regression classifier over a file of fasta sequences, install S-Peaker with the following steps.
1. Download and install the l1_logreg application.
Binaries are available for Intel/AMD Linux, Intel MAC OS X, and powerpc MAC OS X. Alternatively, a source package is made available for compilation on linux systems; Windows compilations are possible but not currently supported by the site. After the installation, make sure that the program l1_logreg_classify can be called from any location in your directory system (by linking to the program from ~/bin or setting the PATH environment variable appropriately, as in this l1_logreg installation example).
2. Create a convenient run directory (e.g.
3. Make sure Java is installed on your system, then download and decompress the LogLikScanner package into your run directory (e.g.
tar -xvf LogLikScanner.tgz).
4. Place the Annotation-Supported-Model directory into your run directory and decompress. (e.g.
tar -xvf Annotation-Supported-Model.tgz).
5. Place your own fasta file (e.g. myseqs.fa) into the Annotation-Supported-Model directory.
6. From the model directory, call the script
speaker.sh on the fasta file to generate features, predict outcome, and create a UCSC custom track for viewing (a complete test example is provided, with fasta sequences and sample calling script