Data Analysis Software for Neurobiology using Matlab

Introduction

As part of an effort to make inexpensive and flexible biology lab instrumentation, we have written a Matlab program which:

• Computes a multi-amplitude windowed spike interval histogram from stored data.
• Computes instantaneous frequency (1/ISI) versus time for each amplitude window.
• Exports the frequency vis time plot to an Excal spreadsheet.

We used Matlab because it is a flexible programming language which includes the abilities to:

• generate graphic user interfaces (GUI).
• acquire data in realtime.
• provide extensive, easy to use analysis and plotting.

Each program description is divided into two parts; program use and program internal structure. Obviously, you don't need to understand the program internals to use it, as long as nothing goes wrong.

Using the Program

If you want to use the program you must have a licensed copy of Matlab release 12. Later in the fall we expect to have a version which will run standalone without Matlab. You then need to download the program we wrote. Be sure that your Matlab 'path' includes the folder where you stored the program. If you are not sure use the SetPath... option in the File menu to investigate. This program is cycle and memory hungary. It may not work well on slow machines. I have tested on an 866 MHz P3 with 256 Meg of memory.

The GUI consists of three plotting axes and their related controls.

The three plots:

• Raw data (from the aqusition program) is plotted at the top.
  • The only control which is active initially is the Load button. Use the dialog box which pops up to select a file saved from the data aqusition program. A simulated test file is here. Cllick the save link as ... option to use this file. The test file contains two simulated units. The small amplitude unit fires about twice as often as the bigger unit. About 200 spikes werre expected for the smaller unit and about 100 for the larger unit. The 7 larger units are summated artifacts.
  • Once data is loaded, you can pick a channel to analyse.
  • Expand T and Reset T buttons enable zooming and panning. Zooming is performed by click-dragging the mouse over the region of interest. Panning is performed by a slider which appears when time is expanded.
  • The Invert button flips the input data.
  • There are two slider controls below the raw data window. The left control sets a beginning analysis cutoff time. The right control sets an ending analysis cutoff time.
  • The Audio button plays the wavefrom through the sound card.
• When you click the Compute button, an amplitude histogram of the spikes is plotted at the lower left and a group of windowed spike interval histograms are plotted at the lower right. A separate histogram is plotted for each active amplitude window.A legend is added to the lower-right plot which gives the averate ISI and number of action potentials in each active amplitude window.
• The number of amplitude windows chosen activates the corresponding number of window threshold buttons. When a window threshold button is selected, the window's lower edge amplitude value may be set. The top edge of the highest window is the maximum amplitude spike detected.
• If you want more resolution in either voltage or interval on either histogram, you can change the number of bins. The Matlab figure zoom control allows you to expand a portion of any plot. The lower-right plot in the example at the top of the page has been zoomed for clarity.
• The 1/ISI vs t button opens a separate window with a plot of 'instantaneous frequency' vs time for each of the windowed spike trains. You have to push the Compute button before you use this feature.
• The 1/ISI-XL button exports the instantaneous frequency vs time data to an Excel spreadsheet. You have to push the Compute button before you use this feature.

Program Internal Organization

The program is structured as a Matlab function. Persistant data is stored in the UserData area of the figure. The initial execution of the function from the command line initializes a bunch of variables and the GUI, then exits. At that point, the GUI is drawn in the figure window and all the GUI controls are waiting patiently for a user action. Control returns to the main function when the user triggers a GUI element, and hence a callback function. Each of the callbacks passes a string to the main function which is dispatched by a gigantic case statement to perform an action, perhaps modify the persistant data, then exit. Since the main function is not running most of the time, the Matlab command line stays active.