... has rid the world of 15358 sites for 17 users over 56 jobs.
MANUEL is an SVM-based classifier that learns how people edit Multiple Sequence Alignments (protein data only). With this knowledge, it performs a manual-quality editing of MSAs and remove invalid sites in a reproducible manner.
Registered users can now submit examples of edited alignments (along with the original alignments) If adding a given example improves the cross-validated accuracy of the classification, it will be added to the default SVM model and credit for the contribution will be given to you.
Citing MANUEL can be done with the following item:
Blouin C., Perry S., Lavell A., Susko E., Roger A. Reproducing the manual annotation of multiple sequence alignments using a SVM classifier, submitted.
The current version of the training set is 1.0. The annotation can be browsed here.
All MSAs from the BaliBASE benchmark (as of Sept 1 2009) were edited by MANUEL. Your may browse the results here.
MANUEL is named after Manuel the waiter from Barcelona in Fawlty Tower. Unlike Manuel, MANUEL should perform exactly what it is intended to do!