add some missing scripts

Author: dkoes

Diff for: calccenters.py

@@ -0,0 +1,50 @@
+#!/usr/bin/env python3
+
+'''Glob through files in current directory looking for */*_ligand.sdf and */*.gninatypes (assuming PDBbind layout).
+Calculate the distance between centers.  If types files are passed, create versions with this information,
+optionally filtering. 
+'''
+
+import sys,glob,argparse,os
+import numpy as np
+import pybel
+import struct
+import openbabel
+
+openbabel.obErrorLog.StopLogging()
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument('typefiles',metavar='file',type=str, nargs='+',help='Types files to process')
+parser.add_argument('--filter',type=float,default=100.0,help='Filter out examples greater the specified value')
+parser.add_argument('--suffix',type=str,default='_wc',help='Suffix for new types files')
+args = parser.parse_args()
+
+centerinfo = dict()
+#first process all gninatypes files in current directory tree
+for ligfile in glob.glob('*/*_ligand.sdf'):
+    mol = next(pybel.readfile('sdf',ligfile))
+    #calc center
+    center = np.mean([a.coords for a in mol.atoms],axis=0)
+    dir = ligfile.split('/')[0]
+    for gtypes in glob.glob('%s/*.gninatypes'%dir):
+        buf = open(gtypes,'rb').read()
+        n = len(buf)/4
+        vals = np.array(struct.unpack('f'*n,buf)).reshape(n/4,4)
+        lcenter = np.mean(vals,axis=0)[0:3]
+        dist = np.linalg.norm(center-lcenter)
+        centerinfo[gtypes] = dist
+
+for tfile in args.typefiles:
+    fname,ext = os.path.splitext(tfile)
+    outname = fname+args.suffix+ext
+    out = open(outname,'w')
+    for line in open(tfile):
+        lfile = line.split('#')[0].split()[-1]
+        if lfile not in centerinfo:
+            print("Missing",lfile,tfile)
+            sys.exit(0)
+        else:
+            d = centerinfo[lfile]
+            if d < args.filter:
+                out.write(line.rstrip()+" %f\n"%d)        

Diff for: cgo_arrow.py

@@ -0,0 +1,81 @@
+'''
+http://pymolwiki.org/index.php/cgo_arrow
+
+(c) 2013 Thomas Holder, Schrodinger Inc.
+
+License: BSD-2-Clause
+'''
+
+from pymol import cmd, cgo, CmdException
+
+
+def cgo_arrow(atom1='pk1', atom2='pk2', radius=0.5, gap=0.0, hlength=-1, hradius=-1,
+              color='blue red', name=''):
+    '''
+DESCRIPTION
+
+    Create a CGO arrow between two picked atoms.
+
+ARGUMENTS
+
+    atom1 = string: single atom selection or list of 3 floats {default: pk1}
+
+    atom2 = string: single atom selection or list of 3 floats {default: pk2}
+
+    radius = float: arrow radius {default: 0.5}
+
+    gap = float: gap between arrow tips and the two atoms {default: 0.0}
+
+    hlength = float: length of head
+
+    hradius = float: radius of head
+
+    color = string: one or two color names {default: blue red}
+
+    name = string: name of CGO object
+    '''
+    from chempy import cpv
+
+    radius, gap = float(radius), float(gap)
+    hlength, hradius = float(hlength), float(hradius)
+
+    try:
+        color1, color2 = color.split()
+    except:
+        color1 = color2 = color
+    color1 = list(cmd.get_color_tuple(color1))
+    color2 = list(cmd.get_color_tuple(color2))
+
+    def get_coord(v):
+        if not isinstance(v, str):
+            return v
+        if v.startswith('['):
+            return cmd.safe_list_eval(v)
+        return cmd.get_atom_coords(v)
+
+    xyz1 = get_coord(atom1)
+    xyz2 = get_coord(atom2)
+    normal = cpv.normalize(cpv.sub(xyz1, xyz2))
+
+    if hlength < 0:
+        hlength = radius * 3.0
+    if hradius < 0:
+        hradius = hlength * 0.6
+
+    if gap:
+        diff = cpv.scale(normal, gap)
+        xyz1 = cpv.sub(xyz1, diff)
+        xyz2 = cpv.add(xyz2, diff)
+
+    xyz3 = cpv.add(cpv.scale(normal, hlength), xyz2)
+
+    obj = [cgo.CYLINDER] + xyz1 + xyz3 + [radius] + color1 + color2 + \
+          [cgo.CONE] + xyz3 + xyz2 + [hradius, 0.0] + color2 + color2 + \
+          [1.0, 0.0]
+
+    if not name:
+        name = cmd.get_unused_name('arrow')
+
+    cmd.load_cgo(obj, name)
+
+cmd.extend('cgo_arrow', cgo_arrow)

Diff for: train.py

@@ -588,7 +588,7 @@ def update_from_result(name, test, result):
 
                 #check if we improved on the test set, if so write a snapshot
                 bests['test_rmsd'], _ , to_snap = check_improvement(test_rmsd, args.update_ratio, bests['test_rmsd'], best_train_interval, i, False)
-                if args.keep_best and to_snap:
+                if args.keep_best and to_snap and not keepsnap: #don't write if already written
                     keepsnap = True
                     print("Writing snapshot because rmsd is better")
                     solver.snapshot() #a bit too much - gigabytes of data     
@@ -605,7 +605,7 @@ def update_from_result(name, test, result):
 
                 #checking if test rmsd_rmse has improved
                 bests['test_rmsd_rmse'], _ , to_snap = check_improvement(test_rmsd_rmse, args.update_ratio, bests['test_rmsd_rmse'], best_train_interval, i, False)
-                if args.keep_best and to_snap:
+                if args.keep_best and to_snap and not keepsnap:
                     keepsnap = True
                     print("Writing snapshot because rmsd_rmse is better")
                     solver.snapshot() #a bit too much - gigabytes of data