PRSTRC is a FORTRAN computer code which utilizes the pseudo probabilities
of structure occurrence frequences of Chou and Fasman (J.M.B. 115, 135-175
(1977), Biochemistry 13, 211-245 (1974), updated in 1978) to predict potential
protein secondary structure.  The alpha helix predicting algorithm first
searches for a set of four residues which satisfy a set of helix "seed"
(nucleation) conditions (at least two "alpha helix formers" and a four residue
average above a threshold).  It then walks in both directions propagating
the helix only through the next four residues (between the current residue and
the one to which it would next H-bond) which have an average value above a
given cutoff. The predicted helix is thus terminated on both sides by
tetrapeptides whose averages have dropped below a given cutoff.  This is a
conservative scheme in that the helix is assumed only over residues which can
H-bond in both directions across helix compatible residues. The beta sheet
prediction scheme searches for a nucleation tripeptide whose average is
above a threshold value.  The sheet is then propagated in both directions
until reaching tripeptides with averages below a given cutoff. Since adjacent
beta strands are not necessarily consecutively located along the amino acid
sequence, the beta strand nucleation threshold should be lowered if any strand
is predicted at a higher conservative level.  This allows for one beta strand
to nucleate beta sheet structures among amino acids which are brought adjacent
only by 3-dimensional folding. Beta turns are predicted using the turn
position dependent Chou-Fasman values.  The position dependent sum must then
be above a turn prediction threshold.  Hydropathy and charge distribution and
the results of the above structural predictions are displayed in association
with each residue.
.
The program was originally developed by William Ralph of Northern Michigan
University and Temple F. Smith of Harvard and has been adapted for the
Molecular Biology Computer Research Resource's users by T.F. Smith, Donald
V. Faulkner, and Teresa A. Webster (1986).
.
Le programme original a ete adapte , et 2 types de sorties condensees ont ete
ajoutee.
.
Pour plus de details sur l'algorithme de calcul de prediction, voir l'article
de RALPH et al. CABIOS (1987) 3,3, 211-216.
"A modified Chou and Fasman protein structure algorithm"
.
Le programme actuellement en place est une version preliminaire de celui
decrit dans l'article de CABIOS . En particilier il n'a pas la prediction
des 'omega loops'.