def load_lasagne_adjacency_nn(nn_dir, epoch):
init_data_pkl = nn_dir + "/init_data/init_data.pkl"
params = proc.load_obj(init_data_pkl)
try:
max_struc_start_idx = params["max_struc_start_idx"],
except KeyError:
max_struc_start_idx = None
try:
max_struc_width = params["max_struc_width"],
except KeyError:
max_struc_width = None
X_tr, y_tr, X_te, y_te = proc.load_lasagne_data(
params["gene_len_fname"],
params["gene_seq_fname"],
params["tr_codons_fname"],
params["te_codons_fname"],
params["outputs_fname"],
rel_cod_idxs=params["rel_cod_idxs"],
rel_nt_idxs=params["rel_nt_idxs"],
rel_struc_idxs=params["rel_struc_idxs"],
struc_fname=params["struc_fname"],
max_struc_start_idx=max_struc_start_idx,
max_struc_width=max_struc_width,
filter_pct=params["filter_pct"])
my_nn = lasagnenn.AdjacencyMLP(X_tr,
y_tr,
X_te,
y_te,
name=params["name"],
out_dir=params["out_dir"],
rel_cod_idxs=params["rel_cod_idxs"],
cod_adj_idxs=params["cod_adj_idxs"],
rel_nt_idxs=params["rel_nt_idxs"],
nt_adj_idxs=params["nt_adj_idxs"],
learning_rate=params["learning_rate"],
update_method=params["update_method"],
widths=params["widths"],
nonlinearity=params["nonlinearity"],
input_drop_rate=params["input_drop_rate"],
hidden_drop_rate=params["hidden_drop_rate"],
num_outputs=params["num_outputs"],
momentum=params["momentum"],
batch_size=params["batch_size"],
reloaded=True)
my_nn.unpickle_epoch(epoch)
return my_nn
def make_lasagne_adjacency_nn(name,
expt_dir,
gene_seq_fname,
gene_len_fname,
tr_codons_fname,
te_codons_fname,
outputs_fname,
rel_cod_idxs=[],
cod_adj_idxs=[],
rel_nt_idxs=[],
nt_adj_idxs=[],
nonlinearity="tanh",
widths=[200],
input_drop_rate=0,
hidden_drop_rate=0,
num_outputs=1,
update_method="sgd",
filter_pct=False,
rel_struc_idxs=False,
struc_fname=False,
max_struc_start_idx=None,
max_struc_width=None,
aa_feats=False,
learning_rate=0.01,
momentum=0.9,
batch_size=500):
X_tr, y_tr, X_te, y_te = setup_lasagne_nn(
name,
expt_dir,
gene_seq_fname,
gene_len_fname,
tr_codons_fname,
te_codons_fname,
outputs_fname,
rel_cod_idxs=rel_cod_idxs,
rel_nt_idxs=rel_nt_idxs,
nonlinearity=nonlinearity,
widths=widths,
input_drop_rate=input_drop_rate,
hidden_drop_rate=hidden_drop_rate,
num_outputs=num_outputs,
update_method=update_method,
filter_max=filter_max,
filter_pct=filter_pct,
rel_struc_idxs=rel_struc_idxs,
struc_fname=struc_fname,
max_struc_start_idx=max_struc_start_idx,
max_struc_width=max_struc_width,
aa_feats=aa_feats,
learning_rate=learning_rate,
momentum=momentum,
batch_size=batch_size)
out_dir = expt_dir + "/lasagne_nn"
_, _, _, params = inspect.getargvalues(inspect.currentframe())
proc.pickle_obj(params,
out_dir + "/{0}/init_data/init_data.pkl".format(name))
my_nn = lasagnenn.AdjacencyMLP(X_tr,
y_tr,
X_te,
y_te,
name=name,
out_dir=out_dir,
rel_cod_idxs=rel_cod_idxs,
cod_adj_idxs=cod_adj_idxs,
rel_nt_idxs=rel_nt_idxs,
nt_adj_idxs=nt_adj_idxs,
learning_rate=learning_rate,
update_method=update_method,
widths=widths,
nonlinearity=nonlinearity,
input_drop_rate=input_drop_rate,
hidden_drop_rate=hidden_drop_rate,
num_outputs=num_outputs,
momentum=momentum,
batch_size=batch_size)
return my_nn