Sephiroth is a disulfide connectivity pattern predictor based on evolutionary information retrieved from Multiple Sequence Alignments (MSAs).
The algorithm has been designed by Gabriele Orlando, Daniele Raimondi and Wim Vranken.

The python implementation of the prediction algorithm can be downloaded, along with PDBCYS and SPX datasets described in the paper from its git repository. It is possible to install it by cloning the repository, with the command:

git clone

or by downloading it as compressed archive. A comprehensive documentation file (README.pdf) is present in the repository.


Requirements/Disclaimer: The code is released under GNU GPL and privileges demonstrative purposes and portability; more efficient implementations can be easily written in languages others than python. Better performances using the current implementation can be achieved by running it on Pypy python optimized interpreter.

  • The current implementation requires only Python's standard libraries.
  • We strongly recommend to use HHblits alignments as input. A local instance of HHblits can be downloaded and installed on your local machine from here. If the chosen parameters and the local HHblits executables' paths are properly set at rows 20-33 in, the script can build HHblits alignments on the fly and use them to predict the query sequence.
  • In order to provide the best tool possible, we are very grateful for bug reports!


Bug Fixes:
- Previous versions of required missing libraries. Now fixed.
- Issues with alignments with redundant sequences fixed.
- The multiple sequence alignment of the protein Q7Z360 has been found to be corrupted: only about half of the sequence length was actually present. Now fixed.