FRAGFOLD by David T. Jones

This software and data relates to the following publications:

D.T. Jones (1997) "Successful ab initio prediction of the tertiary structure of
NK-lysin using multiple sequences and recognized supersecondary structural motifs."
Proteins 29(Suppl 1), 185-191.

D.T. Jones (2001) "Predicting novel protein folds by using FRAGFOLD" Proteins
45(Suppl 5), 127-132.

D.T. Jones, L.J. McGuffin (2003) "Assembling novel protein folds from super-secondary
structural fragments" Proteins 53(Suppl 6), 480-485.

D.T. Jones, K. Bryson, A. Coleman, L.J. McGuffin, M.I. Sadowski, J.S. Sodhi,
J.J. Ward (2005) "Prediction of novel and analogous folds using fragment assembly
and fold recognition" Proteins 61(Suppl 7) 143-151.

T. Kosciolek, D.T. Jones (2014) "De Novo Structure Prediction of Globular Proteins
Aided by Sequence Variation-Derived Contacts" PLoS One 9, E92197.


THIS SOFTWARE MAY BE USED FREE OF CHARGE BY NON-COMMERCIAL USERS. SEE LICENSE FILE FOR DETAILED TERMS.

COMMERCIAL USERS WILL NEED TO OBTAIN A SUITABLE COMMERCIAL USAGE LICENSE FROM UCL BUSINESS - PLEASE CONTACT
enquiries AT ebisu.co.uk FOR FURTHER ASSISTANCE BEFORE YOU ATTEMPT TO INSTALL AND USE THE SOFTWARE.


GENERAL NOTE - IF YOU HAVE AN INTERESTING TARGET PROTEIN TO PREDICT AND WOULD LIKE
US TO RUN FRAGFOLD FOR YOU, THEN PLEASE FEEL FREE TO GET IN TOUCH (e-mail d.t.jones AT ucl.ac.uk).


Compiling:
gcc44 -O3 -ffast-math -march=core2 fragfold.c -lm -o fragfold
(ignore warning about using tmpnam)


FRAGFOLD libraries:
Define FDATA_DIR and FTDB_DIR environment variables and point them to data/ folder.
Otherwise, FRAGFOLD assumes library files are in the same location as FRAGFOLD.


Folding:
fragfold T0281.nfpar output.pdb

Running FRAGFOLD once will generate one single model. For structure prediction you
need to generate an ensemble of say 100 models or more and perform some form of
final model selection (e.g. structural clustering or energy-based selection).
The more models you are able to generate, the more accurate the final model is likely to be.
Typically you would use a cluster of machines to generate these models, but if
you are prepared to wait, you can obviously just run FRAGFOLD on a single computer.

Refinement:
fragfold T0281_refine.nfpar output.pdb T0281_initial.pdb