DataJoint U24 - Workflow Miniscope¶

Setup¶

First, let's change directories to find the dj_local_conf file.

In [1]:

import os

# change to the upper level folder to detect dj_local_conf.json
if os.path.basename(os.getcwd()) == "notebooks":
    os.chdir("..")
assert (
    os.path.basename(os.getcwd()) == "workflow-miniscope"
), "Please move to the workflow directory"
# 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

import os # change to the upper level folder to detect dj_local_conf.json if os.path.basename(os.getcwd()) == "notebooks": os.chdir("..") assert ( os.path.basename(os.getcwd()) == "workflow-miniscope" ), "Please move to the workflow directory" # 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

Next, we populate the python namespace with the required schemas

In [2]:

from workflow_miniscope.pipeline import miniscope, trial, event

from workflow_miniscope.pipeline import miniscope, trial, event

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

Trial and Event schemas¶

Tables in the trial and event schemas specify the structure of your experiment, including block, trial and event timing.

• Session has a 1-to-1 mapping with a behavior recording
• A block is a continuous phase of an experiment that contains repeated instances of a condition, or trials.
• Events may occur within or outside of conditions, either instantaneous or continuous.

The diagram below shows (a) the levels of hierarchy and (b) how the bounds may not completely overlap. A block may not fully capture trials and events may occur outside both blocks/trials.

|----------------------------------------------------------------------------|
|-------------------------------- Session ---------------------------------|__
|-------------------------- BehaviorRecording ---------------------------|____
|----- Block 1 -----|______|----- Block 2 -----|______|----- Block 3 -----|___
| trial 1 || trial 2 |____| trial 3 || trial 4 |____| trial 5 |____| trial 6 |
|_|e1|_|e2||e3|_|e4|__|e5|__|e6||e7||e8||e9||e10||e11|____|e12||e13|_________|
|----------------------------------------------------------------------------|

Let's load some example data. The ingest.py script has a series of loaders to help. If you've already run the other notebooks, you might skip ingest_subjects and ingest_sessions.

In [3]:

from workflow_miniscope.ingest import (
    ingest_subjects,
    ingest_sessions,
    ingest_events,
    ingest_alignment,
)

from workflow_miniscope.ingest import ( ingest_subjects, ingest_sessions, ingest_events, ingest_alignment, )

If you've already run previous notebooks, no need to ingest subjects or sessions.

In [4]:

ingest_subjects()
ingest_sessions()
ingest_events()

ingest_subjects() ingest_sessions() ingest_events()

---- Inserting 0 entry(s) into behavior_recording ----

---- Inserting 0 entry(s) into behavior_recording__file ----

---- Inserting 2 entry(s) into _block ----

---- Inserting 2 entry(s) into _block__attribute ----

---- Inserting 0 entry(s) into #trial_type ----

---- Inserting 50 entry(s) into _trial ----

---- Inserting 50 entry(s) into _trial__attribute ----

---- Inserting 50 entry(s) into _block_trial ----

---- Inserting 0 entry(s) into #event_type ----

---- Inserting 77 entry(s) into _event ----

---- Inserting 77 entry(s) into _trial_event ----

We have 50 total trials, either 'stim' or 'ctrl', with start and stop time

In [6]:

trial.Trial() & "subject='subject1'"

trial.Trial() & "subject='subject1'"

Out[6]:

Experimental trials

subject	session_datetime	trial_id trial number (1-based indexing)	trial_type	trial_start_time (second) relative to recording start	trial_stop_time (second) relative to recording start
subject1	2021-01-01 00:00:01	1	stim	0.193	1.057
subject1	2021-01-01 00:00:01	2	ctrl	1.468	2.332
subject1	2021-01-01 00:00:01	3	ctrl	2.662	3.526
subject1	2021-01-01 00:00:01	4	ctrl	3.738	4.602
subject1	2021-01-01 00:00:01	5	stim	4.826	5.69
subject1	2021-01-01 00:00:01	6	stim	5.973	6.837
subject1	2021-01-01 00:00:01	7	ctrl	7.252	8.116
subject1	2021-01-01 00:00:01	8	ctrl	8.462	9.326
subject1	2021-01-01 00:00:01	9	ctrl	9.731	10.595
subject1	2021-01-01 00:00:01	10	stim	11.019	11.883

...

Total: 50

Each trial is paired with one or more events that take place during the trial window.

In [7]:

trial.TrialEvent() & "trial_id<5" & "subject='subject1'"

trial.TrialEvent() & "trial_id<5" & "subject='subject1'"

Out[7]:

subject	session_datetime	trial_id trial number (1-based indexing)	event_type	event_start_time (second) relative to recording start
subject1	2021-01-01 00:00:01	1	left	0.407
subject1	2021-01-01 00:00:01	1	right	0.269
subject1	2021-01-01 00:00:01	2	center	1.611
subject1	2021-01-01 00:00:01	2	center	1.649
subject1	2021-01-01 00:00:01	3	left	2.777
subject1	2021-01-01 00:00:01	3	left	2.935
subject1	2021-01-01 00:00:01	4	right	4.006

Total: 7

Finally, the AlignmentEvent describes the event of interest and the window we'd like to see around it.

In [7]:

ingest_alignment()

ingest_alignment()

---- Inserting 3 entry(s) into alignment_event ----

In [8]:

event.AlignmentEvent()

event.AlignmentEvent()

Out[8]:

time_shift is seconds to shift with respect to (WRT) a variable

alignment_name	alignment_description	alignment_event_type	alignment_time_shift (s) WRT alignment_event_type	start_event_type	start_time_shift (s) WRT start_event_type	end_event_type	end_time_shift (s) WRT end_event_type
center_button		center	0.0	center	-5.0	center	5.0
left_button		left	0.0	left	-5.0	left	5.0
right_button		right	0.0	right	-5.0	right	5.0

Total: 3

Event-aligned trialized calcium activity¶

In [3]:

from workflow_miniscope import analysis

from workflow_miniscope import analysis

Analysis¶

The analysis schema provides example tables to perform event-aligned Calcium activity analysis.

• ActivityAlignmentCondition - a manual table to specify the inputs and condition for the analysis
• ActivityAlignment - a computed table to extract event-aligned Calcium activity (e.g. dF/F, spikes)

Let's start by creating several analyses configuration - i.e. inserting into ActivityAlignmentCondition

In [4]:

miniscope.Activity()

miniscope.Activity()

Out[4]:

inferred neural activity from fluorescence trace - e.g. dff, spikes

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method
subject1	2021-01-01 00:00:01	0	0	0	caiman_deconvolution
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff

Total: 2

We'll isolate the scan of interest with the following key:

In [18]:

activity_key = (
    miniscope.Activity & {"subject": "subject1", "extraction_method": "caiman_dff"}
).fetch1("KEY")

activity_key = ( miniscope.Activity & {"subject": "subject1", "extraction_method": "caiman_dff"} ).fetch1("KEY")

Here, we can see all trials for this scan:

In [19]:

trial.Trial & activity_key

trial.Trial & activity_key

Out[19]:

Experimental trials

subject	session_datetime	trial_id trial number (1-based indexing)	trial_type	trial_start_time (second) relative to recording start	trial_stop_time (second) relative to recording start
subject1	2021-01-01 00:00:01	1	stim	0.193	1.057
subject1	2021-01-01 00:00:01	2	ctrl	1.468	2.332
subject1	2021-01-01 00:00:01	3	ctrl	2.662	3.526
subject1	2021-01-01 00:00:01	4	ctrl	3.738	4.602
subject1	2021-01-01 00:00:01	5	stim	4.826	5.69
subject1	2021-01-01 00:00:01	6	stim	5.973	6.837
subject1	2021-01-01 00:00:01	7	ctrl	7.252	8.116
subject1	2021-01-01 00:00:01	8	ctrl	8.462	9.326
subject1	2021-01-01 00:00:01	9	ctrl	9.731	10.595
subject1	2021-01-01 00:00:01	10	stim	11.019	11.883

...

Total: 50

And highlight a subset based on trial_type

In [20]:

ctrl_trials = trial.Trial & activity_key & 'trial_type = "ctrl"'
ctrl_trials

ctrl_trials = trial.Trial & activity_key & 'trial_type = "ctrl"' ctrl_trials

Out[20]:

Experimental trials

subject	session_datetime	trial_id trial number (1-based indexing)	trial_type	trial_start_time (second) relative to recording start	trial_stop_time (second) relative to recording start
subject1	2021-01-01 00:00:01	2	ctrl	1.468	2.332
subject1	2021-01-01 00:00:01	3	ctrl	2.662	3.526
subject1	2021-01-01 00:00:01	4	ctrl	3.738	4.602
subject1	2021-01-01 00:00:01	7	ctrl	7.252	8.116
subject1	2021-01-01 00:00:01	8	ctrl	8.462	9.326
subject1	2021-01-01 00:00:01	9	ctrl	9.731	10.595
subject1	2021-01-01 00:00:01	13	ctrl	14.427	15.291
subject1	2021-01-01 00:00:01	14	ctrl	15.563	16.427
subject1	2021-01-01 00:00:01	15	ctrl	16.715	17.579
subject1	2021-01-01 00:00:01	16	ctrl	17.706	18.57

...

Total: 23

Here, we target the event of interest with another key:

In [21]:

alignment_key = (event.AlignmentEvent & 'alignment_name = "center_button"').fetch1(
    "KEY"
)
alignment_key

alignment_key = (event.AlignmentEvent & 'alignment_name = "center_button"').fetch1( "KEY" ) alignment_key

Out[21]:

{'alignment_name': 'center_button'}

In [23]:

alignment_condition = {
    **activity_key,
    **alignment_key,
    "trial_condition": "ctrl_center_button",
}
alignment_condition

alignment_condition = { **activity_key, **alignment_key, "trial_condition": "ctrl_center_button", } alignment_condition

Out[23]:

{'subject': 'subject1',
 'session_datetime': datetime.datetime(2021, 1, 1, 0, 0, 1),
 'recording_id': 0,
 'paramset_id': 0,
 'curation_id': 0,
 'extraction_method': 'caiman_dff',
 'alignment_name': 'center_button',
 'trial_condition': 'ctrl_center_button'}

Next, we add this to the ActivityAlignment table in the analysis schema

In [24]:

analysis.ActivityAlignmentCondition.insert1(alignment_condition, skip_duplicates=True)

analysis.ActivityAlignmentCondition.insert1(alignment_condition, skip_duplicates=True)

In [25]:

analysis.ActivityAlignmentCondition()

analysis.ActivityAlignmentCondition()

Out[25]:

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	condition_description	bin_size bin-size (in second) used to compute the PSTH
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button		0.04

Total: 1

Using the projection method, we can generate a table of relevant trials by trial_type and alignment_condition

In [26]:

sample = (
    analysis.ActivityAlignmentCondition * ctrl_trials & alignment_condition
).proj()
sample

sample = ( analysis.ActivityAlignmentCondition * ctrl_trials & alignment_condition ).proj() sample

Out[26]:

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	trial_id trial number (1-based indexing)
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	2
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	3
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	4
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	7
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	8
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	9
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	13
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	14
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	15
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	16

...

Total: 23

And insert these trials into the ActivityAlignmentCondition.Trial part table

In [27]:

analysis.ActivityAlignmentCondition.Trial.insert(sample, skip_duplicates=True)
analysis.ActivityAlignmentCondition.Trial()

analysis.ActivityAlignmentCondition.Trial.insert(sample, skip_duplicates=True) analysis.ActivityAlignmentCondition.Trial()

Out[27]:

Trials (or subset) to compute event-aligned activity

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	trial_id trial number (1-based indexing)
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	2
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	3
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	4
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	7
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	8
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	9
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	13
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	14
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	15
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	16

...

Total: 23

With the steps above, we have create a new alignment condition for analysis, named ctrl_center_button, which specifies:

• an Activity of interest for analysis
• an event of interest to align the Ca+ activity to - center_button
• a set of trials of interest to perform the analysis on - ctrl trials

---

Now, let's create another set with:

• the same Activity of interest for analysis
• an event of interest to align the Ca+ activity to - center_button
• a set of trials of interest to perform the analysis on - stim trials

In [28]:

stim_trials = trial.Trial & activity_key & 'trial_type = "stim"'
alignment_condition = {
    **activity_key,
    **alignment_key,
    "trial_condition": "stim_center_button",
}
analysis.ActivityAlignmentCondition.insert1(alignment_condition, skip_duplicates=True)
analysis.ActivityAlignmentCondition.Trial.insert(
    (analysis.ActivityAlignmentCondition * stim_trials & alignment_condition).proj(),
    skip_duplicates=True,
)

stim_trials = trial.Trial & activity_key & 'trial_type = "stim"' alignment_condition = { **activity_key, **alignment_key, "trial_condition": "stim_center_button", } analysis.ActivityAlignmentCondition.insert1(alignment_condition, skip_duplicates=True) analysis.ActivityAlignmentCondition.Trial.insert( (analysis.ActivityAlignmentCondition * stim_trials & alignment_condition).proj(), skip_duplicates=True, )

Note the two entries in ActivityAlignmentCondition.trial_condition

In [29]:

analysis.ActivityAlignmentCondition()

analysis.ActivityAlignmentCondition()

Out[29]:

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	condition_description	bin_size bin-size (in second) used to compute the PSTH
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button		0.04
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	stim_center_button		0.04

Total: 2

In [30]:

analysis.ActivityAlignmentCondition.Trial & 'trial_condition = "ctrl_center_button"'

analysis.ActivityAlignmentCondition.Trial & 'trial_condition = "ctrl_center_button"'

Out[30]:

Trials (or subset) to compute event-aligned activity

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	trial_id trial number (1-based indexing)
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	2
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	3
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	4
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	7
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	8
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	9
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	13
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	14
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	15
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	16

...

Total: 23

Computation¶

Just like the element itself, we can run computations with populate()

In [4]:

analysis.ActivityAlignment.populate(display_progress=True)

analysis.ActivityAlignment.populate(display_progress=True)

ActivityAlignment: 100%|█████████████████████████████████████████████████████████| 2/2 [00:10<00:00,  5.34s/it]

In [5]:

analysis.ActivityAlignment()

analysis.ActivityAlignment()

Out[5]:

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	aligned_timestamps
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	stim_center_button	=BLOB=

Total: 2

The AlignedTrialActivity part table captures aligned traces fore each alignment and trial condition specified in the master table.

In [7]:

analysis.ActivityAlignment.AlignedTrialActivity()

analysis.ActivityAlignment.AlignedTrialActivity()

Out[7]:

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method	alignment_name	trial_condition user-friendly name of condition	mask_id	fluorescence_channel 0-based indexing	trial_id trial number (1-based indexing)	aligned_trace (s) Calcium activity aligned to the event time
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	2	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	7	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	8	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	9	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	20	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	27	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	37	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	43	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	46	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff	center_button	ctrl_center_button	1	0	50	=BLOB=

...

Total: 2938

Visualization¶

With the plot_aligned_activities function, we can see the density of activity relative to our alignment event. For more information, see the corresponding docstring.

In [8]:

help(analysis.ActivityAlignment().plot_aligned_activities)

help(analysis.ActivityAlignment().plot_aligned_activities)

Help on method plot_aligned_activities in module workflow_miniscope.analysis:

plot_aligned_activities(key, roi, axs=None, title=None) method of workflow_miniscope.analysis.ActivityAlignment instance
    Plot event-aligned Calcium activities for all selected trials, and
        trial-averaged Calcium activity
        e.g. dF/F, neuropil-corrected dF/F, Calcium events, etc.
    :param key: key of ActivityAlignment master table
    :param roi: miniscope segmentation mask
    :param axs: optional definition of axes for plot.
                Default is plt.subplots(2, 1, figsize=(12, 8))
    :param title: Optional title label

For a refresher on the differences between masks, we can browse the imaging.Segmentation.Mask table.

In [10]:

miniscope.Segmentation.Mask & "mask_id<3"

miniscope.Segmentation.Mask & "mask_id<3"

Out[10]:

A mask produced by segmentation.

subject	session_datetime	recording_id	paramset_id	curation_id	mask_id	segmentation_channel 0-based indexing	mask_npix number of pixels in this mask	mask_center_x (pixels) center x coordinate	mask_center_y (pixels) center y coordinate	mask_xpix (pixels) x coordinates	mask_ypix (pixels) y coordinates	mask_weights weights of the mask at the indices above
subject1	2021-01-01 00:00:01	0	0	0	1	0	209	87	8	=BLOB=	=BLOB=	=BLOB=
subject1	2021-01-01 00:00:01	0	0	0	2	0	896	83	240	=BLOB=	=BLOB=	=BLOB=

Total: 2

Then, we can directly compare the stimulus and control conditions relative to center button presses.

In [6]:

miniscope.Activity()

miniscope.Activity()

Out[6]:

inferred neural activity from fluorescence trace - e.g. dff, spikes

subject	session_datetime	recording_id	paramset_id	curation_id	extraction_method
subject1	2021-01-01 00:00:01	0	0	0	caiman_deconvolution
subject1	2021-01-01 00:00:01	0	0	0	caiman_dff

Total: 2

In [7]:

from workflow_miniscope import analysis
from workflow_miniscope.pipeline import miniscope, trial, event

activity_key = (
    miniscope.Activity
    & {"subject": "subject1", "recording_id": 0}
    & "extraction_method='caiman_dff'"
).fetch1("KEY")
alignment_key = (event.AlignmentEvent & 'alignment_name = "center_button"').fetch1(
    "KEY"
)
alignment_condition_ctrl = {
    **activity_key,
    **alignment_key,
    "trial_condition": "ctrl_center_button",
}
alignment_condition_stim = {
    **activity_key,
    **alignment_key,
    "trial_condition": "stim_center_button",
}

from workflow_miniscope import analysis from workflow_miniscope.pipeline import miniscope, trial, event activity_key = ( miniscope.Activity & {"subject": "subject1", "recording_id": 0} & "extraction_method='caiman_dff'" ).fetch1("KEY") alignment_key = (event.AlignmentEvent & 'alignment_name = "center_button"').fetch1( "KEY" ) alignment_condition_ctrl = { **activity_key, **alignment_key, "trial_condition": "ctrl_center_button", } alignment_condition_stim = { **activity_key, **alignment_key, "trial_condition": "stim_center_button", }

In [8]:

analysis.ActivityAlignment().plot_aligned_activities(
    alignment_condition_stim, roi=2, title="Stimulus Center Button"
)
analysis.ActivityAlignment().plot_aligned_activities(
    alignment_condition_ctrl, roi=2, title="Control Center Button"
)

analysis.ActivityAlignment().plot_aligned_activities( alignment_condition_stim, roi=2, title="Stimulus Center Button" ) analysis.ActivityAlignment().plot_aligned_activities( alignment_condition_ctrl, roi=2, title="Control Center Button" )

In [ ]: