Commit 6264104b authored by Gao, Shang's avatar Gao, Shang
Browse files

added argument for residue_start:residue_end to dcd2array

parent 27220ee2

moldyn_example/scripts/conv_vae.py

+10 −7
Original line number Diff line number Diff line
@@ -84,7 +84,8 @@ class conv_variational_autoencoder(object): 
        

        #even shaped filters may cause problems in theano backend

        even_filters = [f for pair in filter_shapes for f in pair if f % 2 == 0]

        if K.image_dim_ordering() == 'th' and len(even_filters) > 0:

        if (K.image_dim_ordering() == 'th' or K.image_dim_ordering() == 'channels_first') \

                and len(even_filters) > 0:

            warnings.warn('Even shaped filters may cause problems in Theano backend')

        

        self.eps_mean = eps_mean
@@ -92,7 +93,7 @@ class conv_variational_autoencoder(object): 
        self.image_size = image_size

        

        #define input layer

        if K.image_dim_ordering() == 'th':

        if K.image_dim_ordering() == 'th' or K.image_dim_ordering() == 'channels_first':

            self.input = Input(shape=(channels,image_size[0],image_size[1]))

        else:

            self.input = Input(shape=(image_size[0],image_size[1],channels))
@@ -140,7 +141,7 @@ class conv_variational_autoencoder(object): 
        

        #dummy model to get image size after encoding convolutions

        self.decode_conv = []

        if K.image_dim_ordering() == 'th':

        if K.image_dim_ordering() == 'th' or K.image_dim_ordering() == 'channels_first':

            dummy_input = np.ones((1,channels,image_size[0],image_size[1]))

        else:

            dummy_input = np.ones((1,image_size[0],image_size[1],channels))
@@ -155,14 +156,14 @@ class conv_variational_autoencoder(object): 
        

        #define deconvolutional decoding layers

        for i in range(1,conv_layers):

            if K.image_dim_ordering() == 'th':

            if K.image_dim_ordering() == 'th' or K.image_dim_ordering() == 'channels_first':

                dummy_input = np.ones((1,channels,image_size[0],image_size[1]))

            else:

                dummy_input = np.ones((1,image_size[0],image_size[1],channels))

            dummy = Model(self.input, self.encode_conv[-i-1])

            conv_size = list(dummy.predict(dummy_input).shape)

            

            if K.image_dim_ordering() == 'th':

            if K.image_dim_ordering() == 'th' or K.image_dim_ordering() == 'channels_first':

                conv_size[1] = feature_maps[-i]

            else:

                conv_size[3] = feature_maps[-i]
@@ -322,6 +323,8 @@ if __name__ == "__main__": 


    import sys

    import os

    import gzip

    from six.moves import cPickle

    import matplotlib.pyplot as plt

    from scipy.stats import norm
@@ -334,9 +337,9 @@ if __name__ == "__main__": 
    channels = 1

    batch_size = 1000

    conv_layers = 4

    feature_maps = [256,256,256,256]

    feature_maps = [64,64,64,64]

    filter_shapes = [(1,3),(1,3),(3,3),(3,3)]

    strides = [(1,1),(1,3),(1,1),(1,1)]

    strides = [(1,1),(2,2),(1,1),(1,1)]

    dense_layers = 1

    dense_neurons = [128]

    dense_dropouts = [0]

moldyn_example/scripts/dcd2array.py

+4 −4
Original line number Diff line number Diff line
@@ -9,8 +9,8 @@ import shutil 


# get directory to dcd files

args = (sys.argv)

if len(args) != 2:

    raise Exception("Usage: python feature_extraction.py <path to dcd files>")

if len(args) != 3:

    raise Exception("Usage: python feature_extraction.py <path to dcd files> <residue_start:residue_end>")

dcd_path = args[1]

if dcd_path[-1] != '/':

    dcd_path += '/'
@@ -23,7 +23,7 @@ if not os.path.exists('./results'): 
files = glob.glob(dcd_path+'*-protein-*.dcd')

n = len(files)

dataset_name = os.path.basename(files[0]).split('-protein-')[0]

file_basename = os.path.basename(files[0])[:-7]

file_basename = os.path.basename(files[0])[:-8]



# process dcd files

k = 0
@@ -37,7 +37,7 @@ for i in range(n): 


    from MDAnalysis.analysis import contacts

    # crude definition of salt bridges as contacts between CA atoms

    CA = "(name CA and resid 1:21)"

    CA = "(name CA and resid %s)" % args[2]

    # reference groups (first frame of the trajectory, but you could also use a

    # separate PDB, eg crystal structure)

    CA0 = u0.select_atoms(CA)