garnier is an implementation of the original Garnier Osguthorpe Robson algorithm (GOR I) for predicting protein secondary structure. It reads an input protein sequence and writes a standard EMBOSS report file with the predicted secondary structure. The Garnier method is not regarded as the most accurate prediction, but is simple to calculate on most workstations.
The '-idc' option sets offsets to the weights applied to the helix and sheet terms used in the GOR algorithm to assign a residue to a secondary structure state. The option has a value from 0 - 6. This gives an index into a set of arrays, dharr[] and dsarr[], which provide 'decision constants' (dch, dcs), which are offsets that are applied to the weights for the helix and sheet (extend) terms. So, idc=0 says don't use the decision constant offsets, and idc=1 to 6 indicates that various combinations of dch,dcs offsets should be used.
|
|
The meaning and use of the parameter 'idc' is currently being investigated. The original author, Bill Pearson writes:
"In their paper, GOR mention that if you know something about the secondary structure content of the protein you are analyzing, you can do better in prediction. "idc" is an index into a set of arrays, dharr[] and dsarr[], which provide "decision constants" (dch, dcs), which are offsets that are applied to the weights for the helix and sheet (extend) terms. So, idc=0 says don't use the decision constant offsets, and idc=1 to 6 indicates that various combinations of dch,dcs offsets should be used. I don't remember what they are, but I must have gotten the values from their paper."
By default garnier writes a 'tagseq' report file.
The Web servers for PHD, DSC, and others are generally preferred.
Do not rely on this (or any other) program alone to make your predictions with. Use several programs and take a consensus of the results.
The 3D structure for the example sequence is known, although the 2D structure elements were not in the SwissProt feature table for release 38 when the test data was extracted.
DSSP shows:
From To Structure
9 13 E beta sheet
21 39 H alpha helix
50 54 E beta sheet
60 72 H alpha helix
78 81 E beta sheet
85 97 H alpha helix
101 104 E beta sheet
117 119 E beta sheet
128 136 H alpha helix
142 148 E beta sheet
151 166 H alpha helix
170 177 E beta sheet
183 196 H alpha helix
200 204 E beta sheet
208 221 H alpha helix
229 231 E beta sheet
236 239 H alpha helix
244 247 H alpha helix
251 254 E beta sheet
263 273 H alpha helix
284 303 H alpha helix
308 315 H alpha helix
320 322 E beta sheet
325 329 E beta sheet
336 337 E beta sheet
341 345 E beta sheet
351 356 E beta sheet
The accuracy of most stand-alone secondary structure prediction programs is not much better than 70% to 80% at best. The GOR I alogorithm is one of the first semi-successful methods, and will probably not predict with much better than about 65% accuracy. Do not rely on this (or any other) program alone to make your predictions with. Use several programs and take a consensus of the results. The Web servers for PHD, DSC, and others are generally preferred.
This application was modified for inclusion in EMBOSS by