bmtool 0.3.4

BMTool

bmtool

A collection of scripts to make developing networks in BMTK easier.

Getting Started

Installation

pip install bmtool

For developers who will be pulling down additional updates to this repository regularly use the following instead.

git clone https://github.com/tjbanks/bmtool
cd bmtool
python setup.py develop

Then download updates (from this directory) with

git pull

Example Use

> cd your_bmtk_model_directory
> bmtool
Usage: bmtool [OPTIONS] COMMAND [ARGS]...

Options:
  --verbose  Verbose printing
  --help     Show this message and exit.

Commands:
  debug
  plot
  util

>  
> bmtool plot 
Usage: bmtool plot [OPTIONS] COMMAND [ARGS]...

Options:
  --config PATH  Configuration file to use, default: "simulation_config.json"
  --no-display   When set there will be no plot displayed, useful for saving
                 plots
  --help         Show this message and exit.

Commands:
  connection  Display information related to neuron connections
  positions   Plot cell positions for a given set of populations
  raster      Plot the spike raster for a given population
  report      Plot the specified report using BMTK's default report plotter
>
> bmtool plot positions

Plotting Configuration

BMTool utilizes the default simulation-config.json file to know which data files built by BMTK to read. to change this, specify the config after the plot command. Eg:

bmtool plot --config simulation-config-23.json [FUNCTION] 

From python or Jupyter

from bmtool import bmplot
bmplot.plot_3d_positions(config="simulation_config.json")

Ploting Connections

All connection tools can be customized by supplying additional arguments.

Options:
  --title TEXT      change the plot's title
  --save-file TEXT  save plot to path supplied
  --sources TEXT    comma separated list of source node types [default:all]
  --targets TEXT    comma separated list of target node types [default:all]
  --sids TEXT       comma separated list of source node identifiers
                    [default:node_type_id]
  --tids TEXT       comma separated list of target node identifiers
                    [default:node_type_id]
  --no-prepend-pop  When set don't prepend the population name to the unique
                    ids [default:False]

--sources and --targets

Are supplied as comma separated lists and corrospond with the population name specified in your model. Eg:

#initialize the networks in build_network.py
net = NetworkBuilder('hippocampus')
exp0net = NetworkBuilder('exp0input')

Default behavior is to plot connections between all populations but you can specify only a few to simplify your plots.

--sids and --tids

Comma separated lists of node identifiers replace the default cell_id automatically given to a cell population by BMTK. Any parameter passed to NetworkBuilder.add_nodes is stored in network .h5 files and can be used to identify cells while connecting or producing plots. Eg:

# Adding nodes in build_network.py
net.add_nodes(N=inpTotal, pop_name='EC',
    positions=p_EC,
    model_type='biophysical',
    model_template='hoc:IzhiCell_EC2',
    morphology='blank.swc'
    )

We could then use the pop_name to alter the output of our connection plots.

bmtool plot connection --sids pop_name --tids pop_name [FUNCTION]

--no-prepend-pop

Default behavior of bmtool is to print the population name before the cell id (or sid/tid) followed by an underscore. Eg: hippocampus_100. By supplying --no-prepend-pop the cell name becomes 100 unless specified otherwise.

All together basic

Using these optional switches we can see the difference in our plot output below.

Command line

bmtool plot connection total

Python or Jupyter Notebook

from bmtool import bmplot
import matplotlib.pyplot as plt

bmplot.connection_matrix(config="simulation_config.json")

All together advanced

bmtool plot connection --sources hippocampus --targets hippocampus --sids pop_name --tids pop_name --no-prepend-pop --title 'Hippocampus Total Connections' total

Python or Jupyter Notebook

from bmtool import bmplot

bmplot.connection_matrix(config="simulation_config.json", sources="hippocampus", targets="hippocampus", sids="pop_name", tids="pop_name", no_prepend_pop=True, title="Hippocampus Total Connections")

Plot Total Connections

To plot the total number of connections between two populations of cells run

Command line

bmtool plot connection total

Python or Jupyter Notebook

from bmtool import bmplot

bmplot.connection_matrix(config="simulation_config.json", sources="hippocampus", targets="hippocampus")

Remember to customize the output using the instructions above.

--synfo

This is an additional flag that can be used in the total connections plot. By default it is set to '0' which plots total connections. If it is specified as '1', it plots the mean and standard deviation number of connections. If it is '2', it plots the .mod files used for that connection type. Finally if it is '3', it plots the parameter file (.json) used for the connection.

Plot Average Convergence/Divergence

To plot the average convergence or divergence of a single cell excute one of the following commands:

Command Line

bmtool plot connection convergence
bmtool plot connection divergence

Add --method (std, min, or max) for additional function

Python or Jupyter Notebook

from bmtool import bmplot

bmplot.convergence_connection_matrix(config="simulation_config.json")
bmplot.divergence_connection_matrix(config="simulation_config.json")

# OR using methods (min,max,std)
bmplot.convergence_connection_matrix(config="simulation_config.json", method="min")

Plot Connection Diagram

To plot a rough sketch of cell type connectivity and the type of synapse used between cells run:

Command Line

bmtool plot connection network-graph

Python or Jupyter Notebook

from bmtool import bmplot

bmplot.plot_network_graph(config="simulation_config.json")

--edge-property is an option available to change the synapse name if supplied to NetworkBuilder.add_edges when building the network. Default: model_template

Edge Property Histograms

To view the distribution of an edge property between cell types run:

Command Line

bmtool plot connection property-histogram-matrix

Python or Jupyter Notebook

from bmtool import bmplot

bmplot.edge_histogram_matrix(config="simulation_config.json")

The following figure was generated using

bmtool plot connection --sources hippocampus --targets hippocampus --sids pop_name --tids pop_name --no-prepend-pop --title 'Synaptic Weight Distribution between Cell Types' property-histogram-matrix

from bmtool import bmplot

bmplot.edge_histogram_matrix(config="simulation_config.json", sources="hippocampus", targets="hippocampus", sids="pop_name", tids="pop_name", no_prepend_pop=True, title="Synaptic Weight Distribution between Cell Types")

By default the property-histogram-matrix looks at the syn_weight value specified in the NetworkBuilder.add_edges function when building your network. You can change this by specifying the --edge-property. Eg:

bmtool plot connection property-histogram-matrix --edge-property [PROPERTY]

Plotting edge values during/after runtime

BMTool is capable of plotting connection properties obtained after runtime from reports. This is useful for synaptic weights that change over time.

First, you must explicitly record the connection property in your simulation_config.json

  "reports": {
    "syn_report": {
      "cells": "hippocampus",
      "variable_name": "W_nmda",
      "module": "netcon_report",
      "sections": "soma",
      "syn_type": "pyr2pyr",
      "file_name": "syns.h5"
    }
  }

Where pyr2pyr is the POINT_PROCESS name for the synapse you're attempting to record, and the variable_name is a RANGE variable listed int the NEURON block of the synapse .mod file.

Once the simulation has been run un the following referencing the report specified above:

bmtool plot connection property-histogram-matrix --edge-property pyr2pyr_w --report output/syns.h5 --time 9999

The --time-compare option can be be used to show the weight distribution change between the specified times. Eg: --time 0 --time-compare 10000

See the BMTK Commit for more details.

Plotting Distance Probability Matrix between cell types

To show the probability of a cell type being connected to another cell type based on distance run:

bmtool plot connection prob

Full summary of options:

> bmtool plot connection prob --help
Usage: bmtool plot connection prob [OPTIONS]

  Probabilities for a connection between given populations. Distance and
  type dependent

Options:
  --axis TEXT  comma separated list of axis to use for distance measure eg:
               x,y,z or x,y
  --bins TEXT  number of bins to separate distances into (resolution) -
               default: 8
  --line       Create a line plot instead of a binned bar plot
  --verbose    Print plot values for use in another script
  --help       Show this message and exit.

A more complete command (used for image above) may look similar to

bmtool plot connection --sources hippocampus --targets hippocampus --no-prepend-pop --sids pop_name --tids pop_name prob --bins 10 --line --verbose

This will plot cells in the hippocampus network, using the pop_name as the cell identifier. There will be 10 bins created to group the cell distances. A line plot will be generated instead of the default bar chart. All values for each plot will be printed to the console due to the verbose flag.

All point_process cell types will be ignored since they do not have physical locations.

Plotting Current Clamp and Spike Train Info

To plot all current clamp info involved in a simulation, use the following command (uses 'simulation_config.json' as default)

bmtool plot --config simulation_config_foo.json iclamp

To plot all spike trains and their target cells,

bmtool plot --config simulation_config_foo.json input

Printing basic cell information involved in a simulation

bmtool plot --config simulation_config_foo.json cells

Simulation Summary

Using previous functions, plots connection probability as a function of distance, total connections, cell information, current clamp information, input spike train information, and a 3D plot of the network if specified.

bmtool plot --config simulation_config_foo.json summary

Cell Tuning

Python/Jupyter

Single Cell Profiler

from bmtool.singlecell import Profiler

#Example usage
profiler = Profiler(template_dir='./components/templates', mechanism_dir='./components/mechanisms/modfiles')
profiler.passive_properties('Cell_Cf')
profiler.fi_curve('Cell_Cf')
profiler.current_injection('Cell_Cf', post_init_function="insert_mechs(123)", inj_amp=300, inj_delay=100)

Single Cell Tuning

From a BMTK Model directory containing a simulation_config.json file:

bmtool util cell tune --builder

For non-BMTK cell tuning:

bmtool util cell --template TemplateFile.hoc --mod-folder ./ tune --builder

FIR Curve plotting

> bmtool util cell fi --help
Usage: bmtool util cell fi [OPTIONS]

  Creates a NEURON GUI window with FI curve and passive properties

Options:
  --title TEXT
  --min-pa INTEGER   Min pA for injection
  --max-pa INTEGER   Max pA for injection
  --increment FLOAT  Increment the injection by [i] pA
  --tstart INTEGER   Injection start time
  --tdur INTEGER     Duration of injection default:1000ms
  --advanced         Interactive dialog to select injection and recording
                     points
  --help             Show this message and exit.

> bmtool util cell fi
? Select a cell:  (Use arrow keys)
 » CA3PyramidalCell
   DGCell
   IzhiCell
   IzhiCell_BC
   IzhiCell_EC
   IzhiCell_EC2
   IzhiCell_EC_BIO
   IzhiCell_EmoExcitatory
   IzhiCell_EmoInhibitory
   IzhiCell_OLM
   IzhiCell_int

VHalf Segregation Module

Based on the Alturki et al. (2016) paper.

Segregate your channel activation for an easier time tuning your cells.

> bmtool util cell vhseg --help

Usage: bmtool util cell vhseg [OPTIONS]

  Alturki et al. (2016) V1/2 Automated Segregation Interface, simplify
  tuning by separating channel activation

Options:
  --title TEXT
  --tstop INTEGER
  --outhoc TEXT         Specify the file you want the modified cell template
                        written to
  --outfolder TEXT      Specify the directory you want the modified cell
                        template and mod files written to (default: _seg)
  --outappend           Append out instead of overwriting (default: False)
  --debug               Print all debug statements
  --fminpa INTEGER      Starting FI Curve amps (default: 0)
  --fmaxpa INTEGER      Ending FI Curve amps (default: 1000)
  --fincrement INTEGER  Increment the FI Curve amps by supplied pA (default:
                        100)
  --infvars TEXT        Specify the inf variables to plot, skips the wizard.
                        (Comma separated, eg: inf_mech,minf_mech2,ninf_mech2)
  --segvars TEXT        Specify the segregation variables to globally set,
                        skips the wizard. (Comma separated, eg:
                        mseg_mech,nseg_mech2)
  --eleak TEXT          Specify the eleak var manually
  --gleak TEXT          Specify the gleak var manually
  --othersec TEXT       Specify other sections that a window should be
                        generated for (Comma separated, eg: dend[0],dend[1])
  --help                Show this message and exit.

Examples

Wizard Mode (Interactive)

> bmtool util cell vhseg

? Select a cell:  CA3PyramidalCell
Using section dend[0]
? Show other sections? (default: No)  Yes
? Select other sections (space bar to select):  done (2 selections)
? Select inf variables to plot (space bar to select):   done (5 selections)
? Select segregation variables [OR VARIABLES YOU WANT TO CHANGE ON ALL SEGMENTS at the same time] (space bar to select):  done (2 selections)

Command Mode (Non-interactive)

bmtool util cell --template CA3PyramidalCell vhseg --othersec dend[0],dend[1] --infvars inf_im --segvars gbar_im --gleak gl_ichan2CA3 --eleak el_ichan2CA3

Example:

Simple models can utilize

bmtool util cell --hoc cell_template.hoc vhsegbuild --build
bmtool util cell --hoc segmented_template.hoc vhsegbuild

ex: https://github.com/tjbanks/two-cell-hco