Skip to content

Latest commit

 

History

History

function

README.md

Structure-function analysis

Outline

functional analysis flowchart Analysis flowchart: solid boxes indicate data, dashed boxes indicate manuscript display items

scripts

  • compute_neurite_length.py: Compute axon/dendrite length using skeletons.
  • compute_degree.py: Compute out/in-degree from the connectivity graph. (This may take several minutes to run.)
python [script_name.py]

The results will be saved in the data folder as csv format.

notebooks

  • structure_function_analysis.ipynb (Figure 7): Analysis testing whether visual response properties (response strength, intermittency, OSi) are correlated with local connectivity.
  • orientation_direction_tuning.ipynb: Compute tuning curves and determine significantly tuned cells with permutation test.
  • spatial_organization.ipynb (Figure S7): Spatial organization of in-connection density in the volume.
  • visualize_response.ipynb: Visualize responses to different stimuli along with the tuning curve.
  • datajoint_usage_example.ipynb: Contains example codes for querying data from the DataJoint database. Refer to below DataJoint database section for instructions. *This notebook is not executable in mybinder.org.

data

  • pyc_pyc_subgraph.csv: Connectivity graph of pyramidal cell subgraph (only including synapses between pyramidal cells).
  • pyc_all_synapse.csv: Connectivity graph of all synapses that's outgoing from pyramidal cell or incoming to pyramidal cell.
  • Stimulus.pkl: Stimulus labels for all scans.
  • Neuron.pkl: List of segment ids of pyramidal cells.
  • Soma.pkl: Soma locations of pyramidal cells.
  • SynDensity.pkl: In/out-synapse/connection density of pyramidal cells.
  • EASETrace.pkl: Calcium traces of pyramidal cells.
  • EASETuning.pkl: Tuning cruves and tuning curve properties of pyramidal cells.

DataJoint database

This public database which contains extracted structural and functional data such as visual responses and connectivity data.


*Refer to SeungLab DataJoint repository for detailed information of the tables in the schema.

Registration

You need to be registered to access the database.
Please fill out the registration form to receive user id and password. *Currently, it is temporarily available without registration for the review process.

Installation

DataJoint for Python requires Python 3.4 or above to function properly.

pip3 install datajoint

For more information, please checkout the DataJoint Tutorials.

Database configuration

  • HOST: datajoint.ninai.org
  • USER: Given after registration
  • PASSWORD: Given after registration

Accessing the database

import datajoint as dj

# Datajoint credentials
dj.config["database.host"] = "datajoint.ninai.org"
dj.conn() # Then it will ask for your id and password

pinky = dj.create_virtual_module("seung_pinky", "seung_pinky")

*Pinky is the nickname for this dataset named after the American animated television series, Pinky and the Brain.

Fetching data from the database

Example extracting visual response trace from the database.

key = {"segment_id": 648518346349539895, "scan_id": 2}
trace = (pinky.EASETrace() & key).fetch1("trace")

datajoint_usage_examples.ipynb contains examples of using the database to fetch data. Please refer to the DataJoint Tutorials for additional help.

Raw data

2-photon calcium imaging data acquired by Baylor College of Medicine.

2-photon structural stack

  • Structural stack can be downloaded from here.
  • The file contains an array of 512 (h) x 512 (w) x 310 (depth).
  • Voxel dimension is about 0.8 µm x 0.8 µm x 1 µm.
  • The depth of the first section in the stack is assumed as z=0 µm.

Calcium video

  • Calcium videos can be downloaded from here.
  • Each file is a recording of a single slice.
  • Each file is an array of 256 (h) x 256 (w) x 27300 (time).
  • Pixel dimension is about 1.6 µm x 1.6 µm.
  • Frame rate is 14.8313 frames per second.
  • Relative location of the slices (reference to the first section of structural stack).

Visual stimulus

  • Stimulus videos and the labels can be downloaded from here.

Stimulus video

  • There exists one stimulus per scan.
  • Each tiff file is the stimulus video shown for a particular scan.
  • Each tiff file is an array of 90 (h) x 160 (w) x 27300 (time).
  • Frame rate is 14.8313 frames per second.

Stimulus labels

  • Each csv file contains labels of stimulus at each time frame.
  • Each csv file has 27300 values.
  • The value indicates the angle of the directional stimulus at that time frame.
  • If it’s empty (NaN), it indicates the noise stimulus (except for first 200 frames).
  • First 200 frames are black screen stimulus for every scan.
  • Frame rate is 14.8313 frames per second.

Coregistration

  • Coregistration information can be found here.
  • Three different images need to be coregistered: EM, 2p structural stack, and 2p video.
  • 2p structural image is used as the reference. EM and 2p video are coregistered to 2p structural image.
  • Refer to the Colab notebook in the folder to find the transformation.