Overview

AxoGraph is a scientific-analysis and display program for time-series data. Although originally designed for analysis of electrophysiological data, it would be useful to anyone who regularly needs to organize, manipulate, display, and perform curve fitting on data in one or two dimensions. Internally, AxoGraph treats all data as columns of numbers that can be grouped in a variety of ways. But the user is spared having actually to wade through these numbers because the default method of interacting with the data is through an intuitive graphical interface in which the data are presented as "traces" like those on an oscilloscope. The traces can then be grouped, measured, scaled, shuffled, averaged, fitted, or analyzed in more complex ways either using AxoGraph's extensive list of menu-driven analysis routines or user-scripted routines.

Version 4.6 marks a major expansion in functionality with the addition of an optional Data Acquisition Package that allows experiments to be controlled and data acquired directly by AxoGraph (using the supported data-acquisition hardware). The smooth union of analysis and acquisition functions in one program greatly simplifies the practical aspects of running and analyzing experiments, and it provides a high degree of flexibility in experimental design. Versions since 4.0 also have a number of new features including interface improvements and the capability to generate simulations of chemical-reaction kinetics and miniature synaptic currents. Coupled with appropriate acquisition hardware, AxoGraph 4.6 transforms your Macintosh into an all-in-one experimentation and analysis workstation.

How Long Did It Take to Learn to Use It Productively?

Since 1993 I have been using previous versions of AxoGraph as my primary analysis software for patch-clamp data. As I recall, upon first exposure, it took me only a few days to become proficient with almost all of the analysis functionality. Shortly thereafter, I was able to write my own custom single-channel analysis routines using AxoGraph's versatile scripting capabilities. This time around, I had just recently set up a new lab, so testing the acquisition functions was a first-time experience. Within a few hours of installing the software and the acquisition board, I had managed to create stimulus protocols and had used them to perform multi-pulse and multi-episode voltage- and current-clamp experiments on neurons in brain slices. By the end of the day, I had analyzed the data by appropriately grouping, averaging, and measuring the neuronal responses, performed exponential curve fitting on the relevant traces, and generated a publication-quality figure showing the results of the experiment, with fits superimposed.

Overall, Axograph is an extremely user-friendly application. Students doing rotations in my lab, with no prior electrophysiology experience whatsoever, have been able to use it properly within hours. It is entirely menu- and pushbutton-driven, so if you want to do something but don't know how, a few moments of scanning the menu options is usually all that is necessary to find the right function.

As mentioned, the default method of loading, displaying, manipulating, and measuring the data is through graphical displays that resemble oscilloscope traces or other 2-D graphs. For those of us working with time-series data collected from experimental instruments, this is the most intuitive interface possible because it preserves the original form in which the data were first presented to our senses. As with an oscilloscope, one can zoom in or out on regions of interest, using the mouse. In addition to the logically organized menu items, there are several shortcuts to specific functions provided by toolbar icons and hot keys. AxoGraph looks and acts like a Macintosh application should.

Product Quality

Customizability

Along with the data windows, Axograph also opens a text window at startup that serves several useful functions. First, most analysis and acquisition functions both write numerical output to this window and draw it in a data window. The text window is thus a running log that keeps track of what the user has done, and it can be saved for later reference along with any comments about the experiment or analysis. Second, this text window is also an input terminal for AxoGraph's script interpreter, which allows programming commands to be typed and immediately executed. Custom functions can thus be implemented on the fly during an experiment. It can also be used as a simple but very powerful calculator. These scripts can be saved, reloaded, and executed on demand.

Ability to Program in Scripts, Add Extension Modules, etc.

Axograph is fully programmable. There are built-in interpreters for the C, Fortran, Pascal, and BASIC languages, and also capabilities for calling externally compiled programs. Homegrown programs can be added to the menu bar and run at will. Most users will not need these capabilities, but for those interested in pushing the envelope a bit, AxoGraph is up to the challenge.

Ability to Import and Export in Different File Formats

Axograph is able to import ASCII and binary data in a variety of formats, including formats from several other commercial programs such as pClamp, AxoData, AxoScope, Igor , CED, MacLab Chart, and Kaleidagraph. It can export data in ASCII, Igor, Kaleidagraph, or CricketGraph formats, and graphics as PICT files. One can also copy and paste vector graphics or text directly into drawing or word-processing programs.

Useful or Unusual Features

Axograph has a number of features that I find indispensable. Time-series data can come in two basic flavors, episodic or continuous. Episodic data can have a fairly complex structure. For example, an experiment that involves recording responses to a family of 10 different stimuli, where the entire family is repeated 10 times, would result in 100 traces, where every 10th trace is similar but each sequential trace is different. One might wish to examine these data from a number of viewpoints, such as by computing the average and standard deviation of the ensembles formed by grouping every 10th trace, or alternatively by grouping every 10 sequential traces together to see how the fitted time constant of the response varied as the stimulus was varied. Axograph allows you to implement easily any arbitrary grouping and graphing scheme that you wish and to perform the desired analysis on each group. The grouping can then be changed and the data reanalyzed according to a different scheme. Measurements or curve fitting can be applied to every trace within a group, and the results written to the text window can be graphed immediately. This ease and flexibility in graphically organizing data is the basic design philosophy that makes AxoGraph such a valuable tool.

AxoGraph also has special functions for dealing with continuous data, such as spontaneous random events recorded over a long period of time in the presence of noise. The usually excruciating chore of detecting, collecting, and measuring such events is handled smoothly by the plug-in Event Detection Package (included with the distribution). A variety of detection criteria can be applied (e.g., amplitude threshold, or template matching) to tailor the sensitivity of detection for various types of events that might be encountered. By choosing appropriate criteria, therefore, one can detect spikes, waves, exponentially decaying transients, or square steps of any amplitude or polarity, while rejecting noise. The detected events are highlighted during the analysis, allowing the user to make sure that the algorithm is performing as desired. Several measures including event latency, rise-time, amplitude, and half-width can be computed automatically and saved to a separate graphical window as an event log. The event waveforms themselves can be clipped and copied to a new window, generating an ensemble upon which further analysis can be performed. Once the detection parameters have been chosen, the algorithm can perform the detection, measurement, event capture, and display quite rapidly. A signal consisting of hundreds of thousands of data points can be analyzed in a few minutes. The results can immediately be subjected to secondary analysis, such as the generation and fitting of amplitude, rise-time, or inter-event interval histograms.

Finally, the stimulus protocols and the acquired data are stored in similar formats that can be interconverted easily. This means that the acquired data can itself be used as a stimulus. The following anecdote will help to illustrate why I am excited about this functionality: One afternoon, soon after first starting to use AxoGraph 4.6, I created a waveform with desireable statistical properties to be used as a stimulus for studying the input-dependence of neuronal action potential firing. I imported it as ASCII data into AxoGraph, converted it into the stimulus protocol format, delivered it to a neuron as a current-clamp injection, and acquired the resulting neuronal spiketrain. I then immediately took the newly recorded spiketrain, converted that to a stimulus protocol, delivered it as a voltage-clamp command, and recorded the resulting membrane current. This kind of flexibility is almost unprecedented in commercial electrophysiology software and allows one to implement rather complicated experiments almost as soon as they have been conceived.

Limitations

AxoGraph is not a general purpose signal-processing program. Although it has useful built-in functions such as computing power spectra and cross correlations, it does not have the signal-processing functionality of, say, the Matlab Signal Processing Toolbox (although one could probably script these functions if necessary). It is, instead, ideally suited for the sorts of tasks that cellular physiologists routinely require.

There are a few irritating glitches involving memory management and the interface. Occasionally, an analysis function fails for some reason, and a rather cryptic error message is displayed. I believe this may be due to occasional failure of AxoGraph to clear variables properly from memory, because it can often be fixed by restarting the program. Occasionally, the Mac OS Finder gives error messages about insufficient memory, even when nothing but AxoGraph is running, and there is plenty of unused RAM. The data-acquisition interface also has some loose ends that sometimes make it difficult to import and convert custom stimulus waveforms. To be fair, AxoGraph is under continuous development, and judging by the version history, it is likely that these minor concerns will be fixed in future versions.

Comparisons with Similar Software

I have yet to encounter a better analysis tool for electrophysiology, or a better acquisition environment, on any platform. Compared to previous versions, 4.6 has many enhancements but no loss of previously useful functions that I have noticed, and it also offers a considerable speedup when drawing large data sets to the screen. The new acquisition functionality, however, is a major advance that almost seamlessly integrates the many aspects of experimental studies into a single multi-functional application.

Technical Support and Documentation

The documentation is clearly written, includes sections on the technical details of the analysis algorithms, and covers every acquisition and analysis feature. The manual comes as a stand-alone application, and most of the same information can be accessed directly from an online help menu.

Target Users

If you work in the field of cellular electrophysiology and have ever been annoyed by your current acquisition and analysis software, AxoGraph may well be the answer to your prayers. My lab is entirely Macintosh based, and AxoGraph is the core application that I use for almost everything directly related to generating and analyzing data. AxoGraph would also be a great program to use for analyzing any data in the form of columns of numerical values that need to be plotted against each other as line graphs or scatter plots and fitted with linear, exponential, or any other general functions. In summary, if you regularly need to graph, detect, measure, average, scale, filter, or fit your data, you should definitely take a look at this remarkable program.