Palmer Penguins Dataset, analysis using pandas#

We will be using a free Palmer Penguins dataset, publicly available here, to do some statistical analysis.

• First, download the Palmer Penguins dataset to your laptop.

• Log in to an OpenOnDemand instance we use for training. Navigate to File Manager, and uploade the penguins.csv file. Put it right into the home directory.

• In the OpenOnDemand, go to Interactive Apps and find JupyterLab . Start a session on a CPU (no GPUs!) partition using 1-2 CPU cores and 1h of time. Connect to the session when it starts.

• In the JupyterLab launcher, start a Python 3.12 (or whichever is the default kernel) notebook.

Now we are ready to import Python packages. We will need pandas , seaborn and plotting packages.

• Type or copy the code below in notebooks cell:

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

• Run the above cell. Did it succed?

In case some import fails, depending on the software stack, either some Modules need to be added, or pip install in a cell would be used to add missing packages

• Try to fix the Python environment as per above note! Restart the kernel if prompted. Module loads may need re-launching the notebook.

Now that we (hopefully) managed to install the dependencies, lets load the data and check its sanity.

• In a new cell, add the following code, and run the cell:

# Load the data (adjust path below as needed)
penguins = pd.read_csv('~/penguins.csv')
print(penguins.head())
print(penguins.info())
print(penguins['species'].value_counts())

Lets do our “data analysis” and produce some artifact (an image with a plot).

• In a new cell, add the code below and run it

# A visualization: Bill length vs depth by species
plt.figure(figsize=(8, 6))
sns.scatterplot(data=penguins, x='bill_length_mm', y='bill_depth_mm',
hue='species', style='species', s=60)
plt.title('Palmer Penguins: Bill Length vs Depth')
plt.savefig('penguin_scatter.png', dpi=150, bbox_inches='tight')
plt.show()
# Another view: Flipper length vs body mass
plt.figure(figsize=(8, 6))
sns.scatterplot(data=penguins, x='flipper_length_mm', y='body_mass_g',
hue='species', style='sex', s=60)
plt.title('Flipper Length vs Body Mass by Species and Sex')
plt.savefig('penguin_flipper.png', dpi=150, bbox_inches='tight')
plt.show()

At this point, we should have generated two .png files somewhere. Those represents our “results”.

• Locate the “results” files as per above. Use the FIleManager in another browser tab to Download it to your laptop.

• Terminate the JupyterLab session.

Is just closing the tab enough to terminate the job?

Neurodesktop in OOD Desktop#

Neurodesk is a suite of neuro-imaging tools. It is available as Singularity/Apptainer or via CVMFS

• Neurodesk link Apptainer
• Neurodesk link CVMFS

Obtain the brain dataset:#

In an OOD shell or virtual desktop, use _wget to obtain sample brain images.

wget -O sample_T1.nii.gz https://github.com/neurolabusc/niivue-images/raw/main/chris_t1.nii.gz

Find the container using CVMFS#

Alliance systems and Grex include Neurodesk’s CVMFS automatically , no need to follow installation instructions. Lets try to find the images:

ls /cvmfs/neurodesk.ardc.edu.au/containers/ | head

If there is output, we have Neurodesk on CVMFS. Lets try to find latest FSLeyes container there:

#
# finding all images
ls /cvmfs/neurodesk.ardc.edu.au/containers/fsl_*/fsl_*.simg | grep 'fsl_6'
#
# inspecting the last version with singularity
# substitute with apptainer/1.4 for Compute Canada!
module load singularity; 
singularity inspect /cvmfs/neurodesk.ardc.edu.au/containers/fsl_6.0.7.8_20240913/fsl_6.0.7.8_20240913.simg

Now we have found the image . Get an OOD desktop seession (preferably with GPUs!). Open Terminal, load Singularity or Apptainer as per above.

Neurodesk manual suggests using shell command to get command prompt “inside the container”, and then run fsleyes in it.

How to copy/paste into OOD VDI ? Use the NoVNC side panel and Clipboard Icon, then Paste in XFCE Terminal!

module load singularity
singularity shell --nv  /cvmfs/neurodesk.ardc.edu.au/containers/fsl_6.0.7.8_20240913/fsl_6.0.7.8_20240913.simg
# replace the above with Apptainer on CC systems!

In the Apptainer prompt, issue fsleyes command. A GUI should appear. Open the brain image file, and render the results interactively.

Done!#