Electrode Localization¶

Change into the parent directory to find the dj_local_conf.json file.

In [ ]:

# change to the upper level folder to detect dj_local_conf.json
import os
if os.path.basename(os.getcwd()) == "notebooks":
    os.chdir("..")

# change to the upper level folder to detect dj_local_conf.json import os if os.path.basename(os.getcwd()) == "notebooks": os.chdir("..")

In [ ]:

# We'll be working with long tables, so we'll make visualization easier with a limit
import datajoint as dj
dj.config["display.limit"] = 10

# We'll be working with long tables, so we'll make visualization easier with a limit import datajoint as dj dj.config["display.limit"] = 10

Coordinate Framework¶

The Allen Institute hosts brain atlases and ontology trees that we'll use in the next section. The localization.py script assumes this is your first atlas, and that you'll use the 100μm resolution. For finer resolutions, edit voxel_resolution in localization.py. Higher resolution nrrd files are quite large when loaded. Depending on the python environment, the terminal may be killed when loading so much information into memory. To load multiple atlases, increment ccf_id for each unique atlas.

To run this pipeline ...

1. Download the 100μm nrrd and csv files from the links above.
2. Move these files to your ephys root directory.

Next, we'll populate the coordinate framework schema simply by loading it. Because we are loading the whole brain volume, this may take 25 minutes or more.

In [2]:

from workflow_array_ephys.localization import coordinate_framework as ccf

from workflow_array_ephys.localization import coordinate_framework as ccf

Connecting cbroz@dss-db.datajoint.io:3306

In [3]:

dj.Diagram(ccf)

dj.Diagram(ccf)

Out[3]:

Now, to explore the data we just loaded.

In [3]:

ccf.BrainRegionAnnotation.BrainRegion()

ccf.BrainRegionAnnotation.BrainRegion()

Out[3]:

ccf_id CCF ID, a.k.a atlas ID	acronym CHARACTER SET utf8 COLLATE utf8_bin	region_name	region_id	color_code hexcode of the color code of this region
0	6b	Layer 6b isocortex	16	8ADA87
0	a_a_a	Anterior amygdalar area	23	80C0E2
0	a_c_a	Anterior cingulate area	31	40A666
0	a_c_a1	Anterior cingulate area layer 1	572	40A666
0	a_c_a2/3	Anterior cingulate area layer 2/3	1053	40A666
0	a_c_a5	Anterior cingulate area layer 5	739	40A666
0	a_c_a6a	Anterior cingulate area layer 6a	179	40A666
0	a_c_a6b	Anterior cingulate area layer 6b	227	40A666
0	a_c_ad	Anterior cingulate area dorsal part	39	40A666
0	a_c_ad1	Anterior cingulate area dorsal part layer 1	935	40A666

...

Total: 1327

The acronyms listed in the DataJoint table differ slightly from the CCF standard by substituting case-sensitive differences with snake case. To lookup the snake case equivalent, use the retrieve_acronym function.

In [4]:

central_thalamus = ccf.BrainRegionAnnotation.retrieve_acronym("CM")
cranial_nerves = ccf.BrainRegionAnnotation.retrieve_acronym("cm")
print(f"CM: {central_thalamus}\ncm: {cranial_nerves}")

central_thalamus = ccf.BrainRegionAnnotation.retrieve_acronym("CM") cranial_nerves = ccf.BrainRegionAnnotation.retrieve_acronym("cm") print(f"CM: {central_thalamus}\ncm: {cranial_nerves}")

CM: c_m
cm: cm

If your work requires the case-sensitive columns please get in touch with the DataJoint team via StackOverflow.

For this demo, let's look at the dimensions of the central thalamus. To look at other regions, open the CSV you downloaded and search for your desired region.

In [10]:

cm_voxels = ccf.BrainRegionAnnotation.Voxel() & f'acronym="{central_thalamus}"'
cm_voxels

cm_voxels = ccf.BrainRegionAnnotation.Voxel() & f'acronym="{central_thalamus}"' cm_voxels

Out[10]:

ccf_id CCF ID, a.k.a atlas ID	acronym CHARACTER SET utf8 COLLATE utf8_bin	x (um) Anterior-to-Posterior (AP axis)	y (um) Superior-to-Inferior (DV axis)	z (um) Left-to-Right (ML axis)
0	c_m	100	3300	4900
0	c_m	100	3300	5000
0	c_m	100	3300	5100
0	c_m	100	3300	5200
0	c_m	100	3300	5300
0	c_m	100	3300	6100
0	c_m	100	3300	6200
0	c_m	100	3300	6300
0	c_m	100	3300	6400
0	c_m	100	3300	6500

...

Total: 4911

In [17]:

cm_x, cm_y, cm_z = cm_voxels.fetch("x", "y", "z")
print(
    f"The central thalamus extends from \n\tx = {min(cm_x)}  to x = {max(cm_x)}\n\t"
    + f"y = {min(cm_y)} to y = {max(cm_y)}\n\tz = {min(cm_z)} to z = {max(cm_z)}"
)

cm_x, cm_y, cm_z = cm_voxels.fetch("x", "y", "z") print( f"The central thalamus extends from \n\tx = {min(cm_x)} to x = {max(cm_x)}\n\t" + f"y = {min(cm_y)} to y = {max(cm_y)}\n\tz = {min(cm_z)} to z = {max(cm_z)}" )

The central thalamus extends from 
	x = 100  to x = 8400
	y = 2600 to y = 7500
	z = 1700 to z = 9700

Electrode Localization¶

If you have channel_location json files for your data, you can look at the position and regions associated with each electrode. Here, we've added an example file to our pre-existing subject6 for demonstration purposes.

In [4]:

from workflow_array_ephys.localization import coordinate_framework as ccf
from workflow_array_ephys.localization import electrode_localization as eloc

from workflow_array_ephys.localization import coordinate_framework as ccf from workflow_array_ephys.localization import electrode_localization as eloc

In [9]:

(dj.Diagram(eloc) + dj.Diagram(ccf) - 1)

(dj.Diagram(eloc) + dj.Diagram(ccf) - 1)

Out[9]:

Because the probe may not be fully inserted, there will be some electrode positions that occur outside the brain. We register these instances with an IntegrityError warning because we're trying to register a coordinate position with no corresponding location in the ccf.CCF.Voxel table. We can silence these warnings by setting the log level before running populate() on the ElectrodePosition table.

In [ ]:

import logging

import logging

In [3]:

logging.getLogger().setLevel(logging.ERROR)  # or logging.INFO

logging.getLogger().setLevel(logging.ERROR) # or logging.INFO

In [ ]:

eloc.ElectrodePosition.populate()

eloc.ElectrodePosition.populate()

By calling the ElectrodePosition table, we can see the keys the populate() method has already processed.

In [6]:

eloc.ElectrodePosition()

eloc.ElectrodePosition()

Out[6]:

subject	session_datetime	insertion_number	ccf_id CCF ID, a.k.a atlas ID
subject5	2018-07-03 20:32:28	1	0
subject5	2018-07-03 20:32:28	2	0
subject6	2021-01-15 11:16:38	0	0

Total: 3

Let's focus on subject5, insertion 1.

In [ ]:

from workflow_array_ephys.pipeline import ephys

from workflow_array_ephys.pipeline import ephys

In [8]:

key = (ephys.EphysRecording & 'subject="subject5"' & "insertion_number=1").fetch1("KEY")
len(eloc.ElectrodePosition.Electrode & key)

key = (ephys.EphysRecording & 'subject="subject5"' & "insertion_number=1").fetch1("KEY") len(eloc.ElectrodePosition.Electrode & key)

Out[8]:

373

With a resolution of 100μm, adjacent electrodes will very likely be in the same region. Let's look at every 38th electrode to sample 10 across the probe.

If you're interested in more electrodes, decrease the number next to the % modulo operator.

In [13]:

electrode_coordinates = (
    eloc.ElectrodePosition.Electrode & "electrode%38=0" & key
).fetch("electrode", "x", "y", "z", as_dict=True)
for e in electrode_coordinates:
    x, y, z = [e[k] for k in ("x", "y", "z")]
    acronym = (ccf.BrainRegionAnnotation.Voxel & f"x={x}" & f"y={y}" & f"z={z}").fetch1(
        "acronym"
    )
    e["region"] = (
        ccf.BrainRegionAnnotation.BrainRegion & f'acronym="{acronym}"'
    ).fetch1("region_name")
    print("Electrode {electrode} (x={x}, y={y}, z={z}) is in {region}".format(**e))

electrode_coordinates = ( eloc.ElectrodePosition.Electrode & "electrode%38=0" & key ).fetch("electrode", "x", "y", "z", as_dict=True) for e in electrode_coordinates: x, y, z = [e[k] for k in ("x", "y", "z")] acronym = (ccf.BrainRegionAnnotation.Voxel & f"x={x}" & f"y={y}" & f"z={z}").fetch1( "acronym" ) e["region"] = ( ccf.BrainRegionAnnotation.BrainRegion & f'acronym="{acronym}"' ).fetch1("region_name") print("Electrode {electrode} (x={x}, y={y}, z={z}) is in {region}".format(**e))

Electrode 0 (x=11900, y=6300, z=6900) is in Primary auditory area
Electrode 38 (x=11900, y=6300, z=6500) is in Primary auditory area
Electrode 76 (x=11800, y=6300, z=6100) is in Declive (VI) Purkinje layer
Electrode 114 (x=11700, y=6400, z=5600) is in Lobules IV-V molecular layer
Electrode 152 (x=11700, y=6400, z=5200) is in Declive (VI) Purkinje layer
Electrode 190 (x=11600, y=6500, z=4700) is in Primary auditory area
Electrode 228 (x=11600, y=6500, z=4300) is in Granular lamina of the cochlear nuclei
Electrode 266 (x=11600, y=6600, z=3900) is in Cuneiform nucleus
Electrode 304 (x=11600, y=6600, z=3500) is in Lateral hypothalamic area
Electrode 342 (x=11600, y=6700, z=3100) is in root
Electrode 380 (x=11600, y=6700, z=2800) is in root

In [7]:

eloc.ElectrodePosition.Electrode()

eloc.ElectrodePosition.Electrode()

Out[7]:

subject	session_datetime	insertion_number	ccf_id CCF ID, a.k.a atlas ID	probe_type e.g. neuropixels_1.0	electrode electrode index, starts at 0	x (um) Anterior-to-Posterior (AP axis)	y (um) Superior-to-Inferior (DV axis)	z (um) Left-to-Right (ML axis)
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	206	11000	4600	4600
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	207	11000	4600	4600
subject6	2021-01-15 11:16:38	0	0	neuropixels 1.0 - 3B	206	11000	4600	4600
subject6	2021-01-15 11:16:38	0	0	neuropixels 1.0 - 3B	207	11000	4600	4600
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	196	11000	4600	4700
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	197	11000	4600	4700
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	198	11000	4600	4700
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	199	11000	4600	4700
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	200	11000	4600	4700
subject5	2018-07-03 20:32:28	2	0	neuropixels 1.0 - 3A	201	11000	4600	4700

...

Total: 1113