Code used for H01 paper
Code used for the H01 paper (A petavoxel fragment of human cerebral cortex reconstructed at nanoscale resolution) is shared below.
Quality checks of acquired raw EM data
https://github.com/lichtman-lab/mSEM_workflow_manager/tree/9f293f31f1f2b2398ec0cd94b1e1b8e497a75fbf
Preprocessing data for valid tissue detection
Fine-scale Realignment
https://github.com/google-research/sofima/tree/45ca4b12ba1fa9777c8c287ccb1b56fafe60b4dc
Flood-Filling Networks
https://github.com/google/ffn/tree/f92852d8a7659125def757f97c18f3730b1f52c4
Preprocessing segmentation data to remove holes and island segments
Skeletonization of segmentation data
https://github.com/seung-lab/kimimaro/tree/bf0de206b2af766a832913b2ea9922ddc8993d90
Correction of merge errors based on soma annotations and subcompartment classifications
UNet architecture used for synapse identification
https://github.com/UdonDa/3D-UNet-PyTorch/tree/484a8cae0a0ac34cbdcb792c3dac6a547f2827d3
Training split synapse merging algorithm
Applying split synapse synapse merging algorithm
MATLAB scripts for analysis
MATLAB scripts listed here can be run by downloading the repository of all MATLAB code from https://storage.googleapis.com/h01_paper_public_files/H01_Matlab_analysis_scripts.zip and executing the relevant script
Computation of volumes occupied by each class of cellular structure in each layer
Compute and analyse layer 5 and 6 triangular neurons’ basal dendrite angles
Compute spine detachment rates
Render projection images using VASTtools
Calculate minimal distances between different cell types and blood vessels
Python scripts for analysis
Python scripts for analysis of the data are available from the h01 github repository https://github.com/ashapsoncoe/h01. Supporting files, in .zip or original format, for individual scripts are included in the repository, or if large, within a publically-available google cloud storage repository, as indicated in the script in question. Several scripts use Google Cloud BigQuery databases which are not publicly available and require a credentials file to access. However, the data contained within these databases is freely available for download from the released data page.
Randomly select segments by point sampling
Obtain split and merge error statistics for proofread neurons
Plot split errors across the z-axis for axons and dendrites
https://github.com/ashapsoncoe/h01/blob/7d9d435d5da7eed6ac01462d605c694ee1d3fef5/plot_splits.py
Plot synapse density and E:I ratio
Compute E:I ratios of synapses onto neurons
Compute cortical layer bounds
Classify all segments by their cortical layer
Train algorithm to identify shaft of skeletonized axons
Apply algoritm to identify axonal shafts and classify synapses as en passant or terminal bouton
Estimation of distributions of distances from synapse to axonal shaft
Sample points around an axon according the distribution of distances of synpases from the axonal shaft
Sample a range of axons based on their strongest connection to a post-synaptic partner
Analyse axonal connection strengths