def main(path_to_config, parameter_name=None):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("gen_topo_db")
# Fetch adjacency matrix and neuron infos according to configured locations/files
adj_matrix, neuron_info = read_input(stage["inputs"])
assert adj_matrix.shape[0] == len(neuron_info), "Neuron info and adjacency matrix have incompatible sizes!"
topo_db_cfg = stage["config"]
# Calculate tribes, i.e. gids of adjacent neurons
tribes = calculate_tribes(adj_matrix, neuron_info)
# Populate DB
if parameter_name is None: # Case 1: generate all columns at once
DB = create_db_with_specified_columns(["tribe", "neuron_info"] + topo_db_cfg["parameters"],
tribes, neuron_info, topo_db_cfg, adj_matrix)
# Write output to where it's meant to go
write_output(DB, stage["outputs"]["database"])
else: # Case 2: Generate one single column at a time
DB = create_db_with_specified_columns([parameter_name], tribes, neuron_info, topo_db_cfg, adj_matrix)
suffix = "." + parameter_name.lower().replace(" ", "_") # Use parameter name as suffix for output file
# Write output to the 'other' directory for later merging
if not os.path.exists(stage["other"]):
os.makedirs(stage["other"])
out_fn = os.path.join(stage["other"], os.path.split(stage["outputs"]["database"])[1]) + suffix
write_output(DB, out_fn)
def main(path_to_config):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("sample_tribes")
db = read_input(stage["inputs"])
tribes = make_all_samples(db, stage["config"]["Champions"])
write_output(tribes, stage["outputs"])
def main(path_to_config):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("split_spikes")
spikes, stims = read_input(stage["inputs"])
split_spikes = execute_split(spikes, stims, stage["config"])
write_output(split_spikes, stage["outputs"])
def main(path_to_config):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("struc_tribe_analysis")
db, tribal_chiefs, tribal_gids = read_input(stage["inputs"])
tribal_values = lookup_parameters(db, tribal_chiefs, tribal_gids,
stage["config"])
write_output(tribal_values, stage["outputs"])
def main(path_to_config, **kwargs):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("count_triads")
tribes, M, info = read_input(stage["inputs"])
overexpression = count_triads_all(tribes, M, info, stage['config'],
**kwargs)
write_output(overexpression, stage["outputs"])
def main(path_to_config, **kwargs):
assert "index" not in kwargs, "Splitting by index not supported for topo_featurization!"
# 1. Evaluate configuration.
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("topological_featurization")
topo_featurization_cfg = stage["config"]
timebin = topo_featurization_cfg["time_bin"]
parameter = topo_featurization_cfg[topo_featurization_cfg["topo_method"]]
# number of time steps per trial
stim_dur = cfg.stage('split_spikes')['config']['stim_duration_ms']
n_t_bins = int(stim_dur / timebin)
t_bins = np.arange(n_t_bins + 1) * timebin
# 2. Read input data
spiketrains, tribal_data, adj_matrix, neuron_info = read_input(
stage["inputs"])
tribes = tribal_data["gids"]
tribes = tribes.filter(**kwargs)
# 3. Analyze.
# Create analysis function, given the spikes and time bins
featurization_func = make_topo_features_for_tribes(
spiketrains, t_bins, parameter, adj_matrix, GidConverter(neuron_info))
# unpool with this function adds the additional condition of "stimulus" (stimulus identifier)
tribes.unpool(
func=featurization_func) # shape of data: t_bins x 1 x trials
# Now put the data into the expected format. First pooling along tribes (index).
features_data = tribes.pool(
["index"],
func=np.hstack) # shape of data: t_bins x tribe_index x trials
# Then pooling along different stimuli
features_data = features_data.pool(["stimulus"],
func=ordered_list,
xy=True)
# Features that have been removed in the filter step need to be added back for the expected format.
# Update: changed how filter works. It no longer destroys the associated dimension
# for k, v in kwargs.items():
# features_data.add_label(k, v)
# The format also needs an "index" condition. We pooled that away, so we just add to what remains index=0
features_data.add_label("index", "0")
# transform writes data into individual hdf5 files and returns their paths.
fn_data = features_data.transform(
["sampling", "specifier", "index"], # data: str (path to .h5 file)
func=make_write_h5(stage['other']),
xy=True)
# There is some additional info about the neuron samples that we want to inherit from the "tribes" structure.
# So we add that info to fn_data
fn_data.extended_map(lambda x, y: x.update(y[0]),
[get_idv_label(tribal_data)])
write_output(TopoData.condition_collection_to_dict(fn_data),
stage["outputs"])
def main(path_to_config, **kwargs):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("manifold_analysis")
spikes, stims, tribal_chiefs, tribal_gids = read_input(stage["inputs"])
if len(kwargs) > 0:
tribal_gids = tribal_gids.filter(**kwargs)
tribal_chiefs = tribal_chiefs.filter(**kwargs)
res_lookup = transform_all(spikes, stims, tribal_chiefs, tribal_gids,
stage["config"], stage["other"])
write_output(res_lookup, stage["outputs"])
def main(path_to_config):
# Read the meta-config file
cfg = config.Config(path_to_config)
# Get configuration related to the current pipeline stage
stage = cfg.stage("struc_tribe_analysis")
topo_db_cfg = cfg.stage("gen_topo_db")["config"]
stage_cfg = stage["config"]
# Fetch adjacency matrix and neuron infos according to configured locations/files
adj_matrix, neuron_info, tribes = read_input(stage["inputs"])
assert adj_matrix.shape[0] == len(
neuron_info
), "Neuron info and adjacency matrix have incompatible sizes!"
param_dict = get_parameter_db_for_samples(tribes, neuron_info, topo_db_cfg,
stage_cfg, adj_matrix)
# Write output to where it's meant to go
write_output(param_dict, stage["outputs"]["struc_parameters_volumetric"])