Move analysis to pre-process data file

rsp_vision/analysis/preprocess_data.py

@@ -1,10 +1,48 @@
+import logging
+
 import allensdk.brain_observatory.dff as dff_module
+import numpy as np
 from allensdk.brain_observatory.r_neuropil import NeuropilSubtract
 
+from rsp_vision.load.load_raw_suite2p_data import read_numpy_output_of_suite2p
+from rsp_vision.load.load_stimulus_info import how_many_days_in_dataset
+
+
+def pipeline_for_processing_raw_suite2p_data(folder_naming, config):
+ n_days = how_many_days_in_dataset(folder_naming)
+ data = read_numpy_output_of_suite2p(folder_naming, n_days)
+ n_roi = len(data[1]["F"])
+
+ # now handling only the case of one day. Need to extend to multiple
+ # days using registers2p
+
+ dff, r = neuropil_subtraction(
+ data[1]["F"],
+ data[1]["Fneu"],
+ )
+
+ len_session = config["len-session"]
+ n_sessions = config["n-sessions"]
+ dff = dff[:, :, np.newaxis]
+ dff = dff.reshape(n_sessions, n_roi, len_session)
+
+ is_cell = data[1]["iscell"]
+ day = {"stimulus": [1] * n_roi}
+
+ return {
+ "f": dff,
+ "is_cell": is_cell,
+ "day": day,
+ "r_neu": r,
+ "imaging": ["suite2p"],
+ "stim": ["TODO"],
+ "trig": ["TODO"],
+ }
+
 
 def neuropil_subtraction(f, f_neu):
- # for every ROI
  # if fluorescence type is allen_df
+ logging.info("Performing neuropil subtraction...")
 
  # use default parameters for all methods
  neuropil_subtraction = NeuropilSubtract()

rsp_vision/load/load_data.py

@@ -5,10 +5,8 @@
 import h5py
 import yaml
 
-from rsp_vision.load.load_raw_suite2p_data import read_numpy_output_of_suite2p
-from rsp_vision.load.load_stimulus_info import (
- check_how_many_sessions_in_dataset,
- how_many_days_in_dataset,
+from rsp_vision.analysis.preprocess_data import (
+ pipeline_for_processing_raw_suite2p_data,
 )
 
 from ..objects.data_raw import DataRaw
@@ -65,10 +63,12 @@ def load_data_from_filename(
  "Only sf_tf analysis is implemented for summary data"
  )
  else:
- n_days = how_many_days_in_dataset(folder_naming)
- check_how_many_sessions_in_dataset(folder_naming, n_days)
- data = read_numpy_output_of_suite2p(folder_naming, n_days)
- return DataRaw(data)
+ logging.info("Loading raw data from suite2p output")
+ data_dict = pipeline_for_processing_raw_suite2p_data(
+ folder_naming, config
+ )
+
+ return DataRaw(data_dict, is_summary_data=False)
 
 
 def read_config_file(config_path: Path) -> dict:

setup_for_demo.py

@@ -29,6 +29,8 @@
  "drift_order": 2,
  "fps_two_photon": 30,
  "fps_tree_photon": 6.74,
+ "len-session": 11400,
+ "n-sessions": 18,
  "spatial_frequencies": [0.01, 0.02, 0.04, 0.08, 0.16, 0.32],
  "temporal_frequencies": [0.5, 1, 2, 4, 8, 16],
  "directions": [0, 45, 90, 135, 180, 225, 270, 315],