Extracellular Recording Lab

NROB61

Week 09: Extracellular Data Analysis Lab

Table of Contents

Background & Goals 2

Cricket Anatomy 2
The Cricket Cercal System 3
Lab Learning Outcomes 3

Protocol
Part #2: Spike Analysis (second week of the lab) 4
Opening Previously Recorded Data Files 4
Spike Sorting Data Files 7
Analyzing Spontaneous Firing Rates 11
Analyzing Stimulus-Evoked Firing Rates 13

This protocol was adapted from Backyard Brains.

 Extracellular Analysis Lab
NROB61

Background & Goals

Over the next two weeks you will be recording and analyzing data from the common house
cricket, Acheta domesticus. You will collect extracellular spiking activity and will learn some
basics of spike analysis to identify evidence of sensory coding.

In the first module during week 1, "Cricket Recording”, you will collect extracellular spiking data
from a cricket leg preparation, which provides easy access to record from sensory neurons. In
the second module during week 2, "Spike Analysis”, you will analyze previously collected neural
recordings.

Cricket Anatomy
Crickets belong to the order Orthoptera and superfamily Grylloidea, a phylogenetic classification
that also includes grasshoppers and katydids. Crickets are mostly nocturnal and are ectotherms
i.e., their body temperature changes according to the surrounding temperature.

The cricket nervous system is segmentally and bilaterally organized, reflecting annelid
ancestors. The cricket possesses two hind legs divided into four segments that include the
coxa, femur, tibia, and the tarsus. These insects also have a pair of prominent external sensory
antenna-like structures called cerci located on the posterior end of the body. In the image below,
the ovipositor that can be seen between the cerci is only found on female crickets.

Note that the tibia has a row of thorns, or leg

barbs, that are touch-sensitive. Each thorn is
innervated by a sensory neuron, which will elicit
action potentials when sufficient force is applied
to the thorn.

Page 2
 Extracellular Analysis Lab
NROB61

The Cricket Cercal System

The cerci are highly specialized organs, covered in mechanosensitive hairs. Afferent input from
sensory cells associated with mechanosensory hairs innervate two interneurons known as the
lateral and medial giant interneurons. The cerci ganglion are part of a complex sensory system
that integrates wind direction and speed to allow the cricket to detect and evade predators. The
cricket’s cercal-to-giant interneuron system consists of a bilaterally symmetric pair of abdominal
appendages, namely the cerci, and a group of a dozen or so bilaterally homologous first order
interneurons.

Lab Learning Outcomes

In this lab you will:

● Observe and read through the steps involved in preparing a cricket for extracellular
recordings
● Perform extracellular recording in the cricket
● Learn how to view and analyze data files using the Spike Recorder software
● Observe principles of sensory coding by investigating the relationship between stimulus
parameters and neural activity
● Design your own single unit electrophysiology experiment
● Detect the difference between electrical noise and real spikes
● Perform spike sorting on data files to help separate action potentials coming from
different neurons
● Measure spontaneous and stimulus-evoked firing rates from neurons

Page 3
 Extracellular Analysis Lab
NROB61

Protocol
Group Roles for this Activity
Throughout the two weeks of this lab, it is recommended (but not required) that all group
members participate in all activities. Especially during Part 2, when we will perform data
analysis. This will be essential to prepare you for later analyses that you will perform to
complete your final lab report.

Note: Throughout this protocol, you’ll see several questions that correspond to those on your
Quercus submission. You are encouraged to discuss these with your assigned group and take
notes on this Worksheet. Each member will be responsible for submitting their final answers on
Quercus. It is up to each member whether to use the group answer or not.

Part 2: Spike Analysis

Q1: Answer the following questions related to the data you collected last week:

a) Very generally, what type of data did you obtain from the extracellular electrode last
week?
b) Would we be able to measure the post-synaptic potential from the mechanosensory leg
barbs if we placed an extracellular electrode in the tibia?

Opening Pre-recorded Data Files

Locate on Quercus Week 09 Practical page a folder called Cricket Data Files. Download a
local copy of the entire folder to your local computer drive, so you have a personal copy to work
from that won’t be over-written. There should be two audio ‘.wav’ files and one ‘.txt’ text file.

To see the electrical discharge from your recording, open Spike Recorder on the lab laptop.

1. Open Spike Recorder. You should see a green line moving across the middle of the
screen. The application defaults into the “live recording” view, and will monitor input from
microphone. If you clap your hands you’ll likely see a waveform signal move across your
screen.

2. Click on the browse button at the top-right corner to open a previously recorded
file for analysis. After the file browser appears, locate where you have saved your
previously recorded cricket data files. This file is a sample of the spontaneous,
background activity recorded from the coxa-femur leg preparation you performed last
week. No stimulus was actively presented to the leg.

Page 4
 Extracellular Analysis Lab
NROB61
The green line is our recording trace from the electrode, continuously plotting the voltage over
time, similar to the oscilloscope view we saw in the Neuronify Lab. Within this continuous trace,
we will try to detect when action potentials (or “spikes”) have occurred.

First we need to orient to what’s contained in this file and the different functions
available.

3. Press the Play button. Spike Recorder will playback the recording for you. This is what
it looked like during the actual experiment, and you will hear the audio playback to hear
the difference in sound between background noise and the “popping” sounds made
during spikes. You should notice that there are some smaller spikes and some larger
spikes contained within this recording.

(Can you hear the difference in sound between the small and large spikes?)

4. Make the signal larger/smaller by adjusting the range of the y-axis. You want to
ensure you can view both the maximum and minimum peaks of your largest spikes,
making sure they are not falling outside the viewable window.

(How many different neurons do you think this electrode is picking up? How can you tell?)

5. Zoom in on the time x-axis by scrolling up or down with your mouse/trackpad, and you
will see an individual spike is all its glory:

Page 5
 Extracellular Analysis Lab
NROB61

Notice that the time scale bar changes as you scroll up and down using your mouse.

Keeping track of the where you are within the recorded file in terms of real “experiment time”
can be a little confusing in this program since time isn’t plotted on a continuous scale and there
is no fixed cursor keeping track of your position. In Spike Recorder, your current time position
within the recorded file is being displayed in the bottom left corner; however, this time position
corresponds to the time point measured at the right-hand edge of the viewing window. The
next steps are to help you understand this better.

6. Drag the time scrollbar all the way to the left-most position at the start of the
recording file. Recording traces will always start with a flat green line, because the
actual start position of the file (Time 00:00 000) is occurring at the right edge of the
window. Data does not exist within the flat line portion. Scroll to the right very slowly to
see the background noise start.

7. Drag the time scrollbar all the way to the right-most position at the end of the
recording file. Note that your current time position changes as you scroll, but your total
file duration time does not. When you get to the end of the file, the current time reading
should now match the total duration of the recording file.

Q2: What is the total duration of this file, in seconds? (round to nearest whole second, using no
decimal points or units)

8. Locate the last large-amplitude spike in the recording file. Zoom in until the time
scale bar reads “0.1 s”. Readjust the scrollbar, if necessary, to confirm you are still
viewing the last large spike. Precisely align the peak of the positive inflection to the right
edge of the window and record the time position within the file.

Tip: You’ll have more fine-control over the position if you click-and-drag the green
recording trace left and right instead of moving the scrollbar. You should also maximize
the Spike Recorder window to your full screen size.

Page 6
 Extracellular Analysis Lab
NROB61

Q3: At what time did the last large-amplitude spike occur? Measure at the peak positive
inflection of the spike. (Enter the timestamp exactly as it is reported in Spike Recorder. Include
all numbers and punctuation)

The most commonly used measure of background noise is the root-mean-square (RMS) of the
data over a given window, and is essentially the standard deviation of the voltage. RMS
measurements can be made after selecting a measurement window.

9. Locate any region of the file where there is no detectable spiking activity for at
least 0.5 s. You can hold the right-click button (or press two fingers on trackpad on Mac
OSX) and drag the mouse cursor to select a measure window. The duration of your
selection will appear towards the bottom. Once it is set at 0.5 s, you can let go of the
cursor and the measurement window will stay. You can move the scrollbar or drag the
voltage trace if you need to adjust where it is placed.

Note: You can test moving this window around to different sections of background noise
in your file. You should get roughly the same measurement throughout.

Q4: What is the approximate RMS value of the background noise?

Spike Sorting Data Files

Once a recording is made, spike thresholding needs to take place as a way of separating neural
signals from background noise. In cases where your electrode happens to pick up activity from
more than one neuron, you will also need to “spike sort” in order to classify which neuron each
spike belongs to.

In Quercus there were two assigned videos on Spike Sorting. The methods described in the
video are a little more sophisticated than we will be doing in this lab, but provides you with a
more realistic idea of what spike sorting involves. Spike sorting is a very labor-intensive
process; for every hour of experimental recordings, researchers spend hours spike sorting their
file. However, advances in automated spike sorting algorithms are available, but less reliable.

For this lab, I have curated some files for you that allow for simpler spike sorting and that can be
done by setting simple amplitude threshold ranges. Within DataFile1, we can presume that

Page 7
 Extracellular Analysis Lab
NROB61
there is spontaneous activity from two different neurons being recorded, but we need to
separately classify each spike now.

10. Open the Spike Sorting Analysis tool by clicking the conical flask button at the
top. This will bring you to a new window view.

The Spike Analysis window showing an example of spike sorting and the different button selections available.
Only the largest-amplitude spikes are being detected between the two blue lines. These spikes are assigned
a unique MarkerID and the panel at the right shows this unit as having 657 spikes total across this recording.

11. You will see a coloured line appear over your recording. To detect spikes belonging to
one neuron, grab the large end of the line and pull it up or down (click on the area
where the “Spikes selection” arrow is pointing in the image above). This will reveal a
second line underneath of the same colour that you can also drag up or down. These
two lines will act as the upper and lower amplitude bounds for detecting spikes
belonging to one neuron; any spikes falling in-between these two lines will be classified
as belonging to the same neuron, and will be labeled with that colour.

You will need to decide whether it’s best to detect spikes using the positive inflection OR the
negative inflection. Do not count the same spike twice. In some cases it won’t matter, but in
other cases it will make a big difference. Let’s see if it makes a difference in this file:

12. Accurately discriminate all of the largest-amplitude spikes as belonging to the

same MarkerID. Try sorting on the positive inflection first. Be careful to avoid
including background noise or any smaller-amplitude spikes that don’t share a similar

Page 8
 Extracellular Analysis Lab
NROB61
spike height. The total number of spikes falling within this classification will be reported
in the panel to the right, under the heading “Channel 1”.

13. Try moving your lines to do the same analysis on the large-amplitude spikes, but
this time setting the threshold lines on the negative inflection instead.

(Was there a difference in spike count between Step 12 and Step 13? If there was,
which spike count would you trust more and why?)

Q5: How many spikes belong to the large-amplitude unit? Make sure you are not counting the
same spike twice.

14. Click on the plus button under “Channel 1” panel to add a new spike MarkerID.
This will bring up a new set of threshold lines with a different colour to identify a second
neuron.

Tip: if you ever want to delete one of your MarkerIDs, just click on the minus sign next to
the spike count.

15. There is a set of smaller-amplitude spikes, with a peak height that is ~25% the height of
the large unit you’ve already classified. Try to capture all of these spikes, setting your
upper and lower threshold lines so that you avoid the background noise as much
as possible and also avoid the large-amplitude unit from being included.

(Do you think it would be better to threshold on the positive or negative inflection for this
smaller-amplitude unit? Why?)

Q6: What is the spike count belonging to the smaller-amplitude unit?

16. Click the Save button. This will create a new “DataFile1-events.txt” text file associated
with this recording file. This text file will now include all of the timestamps of each
detected spike, along with the MarkerID it was assigned to, so that they will re-appear
the next time you load the file. These text files can also be ready by other programs to
create raster plots and PSTHs, or run more sophisticated analyses on the spike trains.

Note: Each time you save it will overwrite and update the events marked in the text file.

Tip: If you receive an error trying to save the text file, you will need to refer to the
Troubleshooting Guide on Quercus.

17. Click on the flask button to return to the main recording window. You will now see
coloured dots above your spikes, indicating the MarkerID they have been assigned.
Scan through your file to decide if you are happy with the results.

Page 9
 Extracellular Analysis Lab
NROB61

Back to the main window, an example of spikes being clearly identified

correctly and a coloured MarkerID dot placed above the spike.

(Are there any spikes that don’t have a dot above them? Any spikes that were double-
counted, with 2 dots above them?)

18. For the smaller-amplitude unit, you might have found instances where you weren’t as
confident about the placement of your discrimination lines. If you are too close to the
background noise, you might detect extra spikes that don’t actually belong (false
positives), but as you move your line away from the background, you might miss spikes
that actually belong to the spike train (false negative).

(Do you think it’s worse to have false positives or false negatives?)

Q7: Can you think of a way that the classifications for these smaller-amplitude spikes might be
improved to reduce false positives/negatives? Briefly explain.

19. Go back to the Spike Analysis window. For demonstration purposes only, click the
plus button again to add a third marker that detects the background noise. Place one line
centered right in the middle of the background noise, and place the second threshold line
just above the background. You should detect ~11.6K spikes.

20. In the bottom left corner, click the arrow to open up the statistical graphs. It should
default to display the first tab option, “Average Waveform”. For every detected spike,
Spike Recorder has automatically taken a 5-ms “snippet” of the recording trace, centered
around the maximum peak of your detected spike (i.e., 2.5-ms on either side). Aligning
all of the peak amplitudes to time zero, the average waveform is calculated, along with 1
standard deviation of variance. Well-isolated single-units will have large amplitudes, low
variance, and you will be able to visualize the biphasic or triphasic shape of the
waveform. Smaller spikes that are very close to the background noise will have low
amplitude, high variance and only show one peak in the average waveform.

Tip: We aren’t usually interested in analyzing the background noise, but it’s helpful for
you to see what this looks like, so you can avoid analyzing data that looks like this in the
future. If you see this waveform shape when analyzing data, re-think your threshold lines
or be aware that you are looking at an averaged response from many, many different

Page 10
 Extracellular Analysis Lab
NROB61
neurons!

Q8: Take a screenshot of your 3 average waveforms, and save the image to upload to
Quercus.

Q9: Which unit has the smallest amount of variance to its waveform?

Q10: Looking at the average waveform plot from the large-amplitude neuron, what is the
maximum amplitude measured at its peak? (Note: the maximum amplitude might occur on the
positive or negative deflection of the spike. Look for the largest change from baseline and read
the measurement off of the y-axis.)

21. Within the statistical graph region, click the tab labelled “Inter-spike Interval”. For
this plot, the program calculates the time difference between adjacent spikes belonging
to the same MarkerID. The x-axis is plotted on a logarithmic time scale to capture both
small and large intervals. For example, -2 on the horizontal axis represents 10-2 seconds,
which is equal to 0.01 s, or 10 ms. The y-axis is frequency of observations (i.e., how
many times adjacent spikes were found to have that particular inter-spike interval within
the recording file).

Q11: Briefly explain how one should interpret the inter-spike interval graph, and how this could
be useful for the spike sorting process. (Hint: what is the minimum inter-spike interval you would
expect to see from a well-isolated individual neuron, and why?)

Analyzing Spontaneous Firing Rates

Once you are happy with your spike sorting and have detected spike events within your file, you
can move on to analyzing spontaneous firing rates. From the main Spike Recorder window
we can specify a region of interest to calculate spike counts and firing rates from, by selecting a
measurement window like we did for calculating RMS.

22. Click and drag with the right mouse button to select a 5 second duration
measurement window. Align the start of the measurement window with start of the
recording file (Time 00:00), where the background noise begins. Add the spike count
and firing rate measurement to the table below.

Note: The raw spike count is the first number that appears next to the coloured
Marked ID. The firing rate is within parenthesis next to the spike count, and
shows the units of “Hz”).

Page 11
 Extracellular Analysis Lab
NROB61

Sample recording showing signal measurements taken after spike sorting completed. Here, a 500ms
window is specified and aligned to the onset of a stimulus presentation. Raw spike counts and firing
rate are calculated for all MarkerIDs (top, left). In the first 500-ms of the stimulus, the purple unit has a
firing rate of 138 Hz, whereas the red unit’s firing rate is only 8 Hz (i.e., 4 spikes/0.5 s = 8 sp/s or 8 Hz).

23. Shift your measurement window over to measure the firing rate during the next 5 s
of the recording file, and repeat.

Tip: Before moving your first measurement window that started at time zero, write down
the current time position of your window (bottom left; recall that this is the timepoint at
the right edge of the screen). Add 5 sec to this value. Without changing the time
resolution, drag the recording trace over to the left, until the new current time position
reads the new value you calculated. See video demo on Quercus if you need more help.

Q12: Record your measurements in the sample table provided below and then calculate the
average and standard deviation across these 5 samples.

Measurement 5-sec Spontaneous 5-sec Spontaneous

Page 12
 Extracellular Analysis Lab
NROB61
s Firing Rate (Hz) Firing Rate (Hz)
[large-amplitude unit] [small-amplitude unit]
#1
#2
#3
#4
#5
Average
Standard
Deviation

Analyzing Stimulus-Evoked Firing Rates

You should now have all of the necessary skills to start analyzing and comparing stimulus-
evoked firing rates to search for evidence of sensory coding! This will be a key aspect to your
Final Lab Report.

It will be critical for the Final Lab Report that everyone is able to use Spike Recorder on their
own computer and can save .txt files of their spike sorts independently. You may discuss this
section with your assigned group, but each member should complete the steps on their own
device and submit their own work.

This last section walks you through one more demonstration of the exact steps needed to
compare stimulus-evoked responses between stimuli.

24. Click on the browse button at the top-right corner, to open “DataFile2.wav” file.

This recording was made while recording from the coxa-femur leg preparation. This file contains
two numbered Event Markers, which appear as a vertical line with a number label at the top.
These can be added during an experiment to mark important time events, such as presenting a
new stimulus. (If you cannot see the Event Markers, refer to the Troubleshooting Guide on
Quercus).

During this experimental recording, the following sequence of events occurred:

 The recording was started, with ~20 sec of spontaneous activity recorded.
 Event 1 (low intensity): 7 presentations of weak air pressure were directed at the tibia
from 10-cm away. Stimulus duration was ~1.5 s, with ~5 s pause in-between each trial.
 Event 2 (high intensity): 7 presentations of strong air pressure were directed at the
tibia from 5-cm away. Stimulus duration was ~1.5 s, with ~5 s pause in-between each
trial.

(Based on your understanding of sensory coding in mechanoreceptors, how do you think

neurons will respond differently to the weak vs strong stimulus intensity? Practice writing a

Page 13
 Extracellular Analysis Lab
NROB61
clear hypothesis statement based on the specific experiment described above and share
with your group members).

For each stimulus presentation, a burst of evoked action potentials can be observed in response
to it, lasting the duration of the air stimulus applied. After Event Marker “1” you should see 7
large spike train bursts, and after Event Marker “2” another set of 7 large response bursts. Since
there was no way to digitally mark exactly when the stimulus was presented for each trial in this
experiment, we will assume that the first spike in each spike train burst also marks stimulus
onset.

25. Spike sort the data file, and click Save. This stores timestamps of every detected
spike based on your thresholding. The next time you open this file, it will remember your
spike sorting results. Return to the main window. Check the quality of your detected
spikes and adjust if necessary.

(How many unique neurons do you think can be sorted in this file? Did you decide to
threshold on the positive or negative deflection, and why?)

Q13: You will be asked to submit your final saved .txt file to Quercus to ensure you are
able to do this step on your own.

26. Move to the file region with Event Marker “1” visible. Set a measurement window
of 1.5 s duration. Align your measurement window so that you can record the
spontaneous (pre-stimulus) firing rate during the 1.5 s immediately preceding the start of
the first stimulus. Record the spontaneous firing rate from your spike sorted unit(s).

27. Grab and drag the green recording trace so that you can align the measurement
window to now start at the first detected spike within the spike train burst. Keep
the measurement window at the same 1.5 s duration, regardless of whether the evoked
response ends before or after the end of the measurement window. Record the stimulus-
evoked firing rate your spike sorted unit(s). This allows us to look for a difference in the
firing rate before and after the stimulus, measured over any identical amount of time.

28. Repeat for all 7 stimulus presentations of low-intensity trials (Event Marker “1”).
Calculate mean firing rates and standard deviation across the 7 trials.

29. Move to the file region after Event Marker “2”. Measure spontaneous and
stimulus-evoked firing rates for the 7 trials of the high-intensity stimulus.

Create a summary table of all your measurements, showing both the spontaneous and stimulus-
evoked firing rates for the low-intensity and high-intensity trials. Calculate the average firing
rates and standard deviations.

Measurements 1.5-sec Spontaneous 1.5-sec Stimulus-Evoked

(Low intensity) Firing Rate (Hz) Firing Rate (Hz)

Page 14
 Extracellular Analysis Lab
NROB61
[unit ID] [unit ID]
#1
#2
#3
#4
#5
#6
#7
Average
Standard
Deviation

Measurements 1.5-sec Spontaneous 1.5-sec Stimulus-Evoked

(High intensity) Firing Rate (Hz) Firing Rate (Hz)
[unit ID] [unit ID]
#1
#2
#3
#4
#5
#6
#7
Average
Standard
Deviation

Tip: you can calculate standard deviation easily within Excel using the “=stdev( )” or
“=stdev.s( )” function, and selecting your 7 trial measurements for the numbers to
include.

Q14: Copy your completed measurement tables into Quercus, for just one sorted neuron.

Note: If you spike-sorted more than one neuron, you would need to repeat the above
tables for each unique neuron IDs. For your submission this week, you only need to turn
in a set of tables corresponding to one isolated neuron.

Q15: Copy a screen-shot image of the average waveforms from any neuron(s) you sorted.

Page 15