************************************************************ Here is an example of the output from I-SITES / ROSETTA Coordinates are PDB format. The B-factor column contains the confidence of the local structure prediction at that point in the sequence. Bond lengths and angles are fixed to their ideal values. The header lines contain some bare-bones explanations of the steps taken in predicting the structure, formatted as a CASP submission. ************************************************************ PFRMAT TS TARGET pred REMARK METHOD Automated I-SITES / ROSETTA METHOD ========================================================= METHOD I-sites predicts local structure motifs in the query METHOD sequence. These are used by Rosetta to generate the METHOD 3D structure by fragment insertion Monte Carlo. METHOD The output predictions are coordinates. The predictions METHOD are ab initio. No fold recognition is done by this server METHOD at the current time. METHOD The predictions from this server are of local and METHOD super-local structure. They should be evaluated METHOD on an approximate length scale of 30-60 residues. METHOD Some effort is made by the program to pack the local structure METHOD into a globally-globular form. METHOD ========================================================= METHOD Steps used to generate TS predictions METHOD (1) Psi-blast was used to obtain an alignment METHOD (2) The PHD prediction server was used to obtain a 3-state METHOD prediction (optional) METHOD (3) The I-sites method (JMB 281(3):565-577) METHOD was used to predict backbone angles and METHOD confidences. Backbone angle confidnces are reflected in METHOD the B-column of the PDB output. B = 1.00 + (1.00 - confidence)*10 METHOD except unpredicted residus (B = 0.00) METHOD (4) Non-conserved non-polar sidechains were converted to their METHOD most likely polar counterpart. This would, we hope, restrict METHOD core residues to conserved non-polar sidechains. METHOD (5) Fragment insertion Monte Carlo folding (ROSETTA, Simons et al METHOD PNAS 1999) was used to build a 3D structure from the METHOD fragment library output by I-sites. Residues with I-sites METHOD confidences > 0.75 were restrained to their predicted backbone METHOD angles. For longer sequences, the protein was broken up into METHOD segments containing 50 un-restrained residues (typically 100 METHOD residues). METHOD (6) The 3D structure with the best score was selected, its METHOD sequence replaced with the original one. For long sequences METHOD A simple genetic algorithm was employed to find the METHOD optimum points of recombination, using the same scoring function METHOD as above (ROSETTA). METHOD METHOD >>>>THIS METHOD IS TOTALLY AUTOMATED!! <<<< METHOD URL http://honduras.bio.rpi.edu/~isites/hmmstr/server.html METHOD MODEL 1 REFINED PARENT N/A ATOM 1 N VAL 69 1.221 -0.771 -1.555 1.00 5.45 ATOM 2 CA VAL 69 0.181 -1.145 -0.603 1.00 5.45 ATOM 3 C VAL 69 0.448 -0.540 0.769 1.00 5.45 ATOM 4 O VAL 69 -0.432 0.080 1.366 1.00 5.45 ATOM 5 CB VAL 69 0.072 -2.659 -0.505 1.00 5.45 ATOM 6 N LEU 70 1.666 -0.726 1.266 1.00 4.80 ATOM 7 CA LEU 70 2.077 -0.129 2.531 1.00 4.80 ATOM 8 C LEU 70 1.675 1.339 2.604 1.00 4.80 ATOM 9 O LEU 70 1.086 1.783 3.589 1.00 4.80 ATOM 10 CB LEU 70 3.577 -0.281 2.727 1.00 4.80 ATOM 11 N THR 71 2.000 2.089 1.556 1.00 4.50 ATOM 12 CA THR 71 1.631 3.497 1.477 1.00 4.50