Lunarc Aurora

Hints

Imaging data visualisation

  1. connect to LUNARC using ThinLinc
  2. upload your data to your nobackup directory /lunarc/nobackup/users/'your user name'

ImageJ

Applications -> Lunarc Applications -> Medical Imaging -> fiji

Tomviz

Applications -> Lunarc Applications -> Volume Rendering -> Tomviz

Slicer3D

Applications -> Lunarc Applications -> Volume Rendering -> Slicer3D

Running Jupyter notebook on LUNARC with small data (algorithm development)

Do the following if you want to tweak a script on a small dataset ( e.g. 300 x 300 x 300 pixels):

module purge
module load Anaconda3
#start a new jupyter notebook:
jupyter notebook

#if you have an example notebook you can load it directly by navigating to the directory where your script is. Easiest is if you navigate to the directory where your data is and put the python script there
cd /nobackup/YourData
jupyter notebook 'notebookName.ipynb'

#if you want for example to download a notebook from the QIM homepage for example, open it in the web browser in LUNARC and download the *.ipynb into your /nobackup directory. Then follow the abowe instructions in this paragraph

Using SPYDER editor of Python scripts on LUNARC

  1. connected to LUNARC using ThinLinc
  2. go to Applications -> Lunarc Applications -> Python -> Spyder - Anaconda3

Loading image data, cropping, downscaling

Here we try to provide scirpts for loading in data and handling them before you start the analysis / visualisation

  1. One example how to load in a stack of tiff files and downscale it you find here: Kevin's kaggle script

Image Quantification

  1. see examples at QIM

Getting exclusive node with jupyter-notebook

Lunarc is working on the implementation of a proper JupyterHub but it is not available for the time of hackathon. So we suggest to use a trick you can find on the Web, e.g. here: Remote Access to IPython Notebooks via SSH. The trick is: 1. allocating the node, getting its hostname and starting the jupyter notobook 2. making a ssh tunnel from the frontend to the given node 3. connecting to the jupyter notebook from a browser at the frontend

# 1) open a terminal and get the node:
interactive -A lu2018-2-48 -p lu --mem=210GB -t 12:00:00 --exclusive

# 2) type there "hostname" in order to get node name (e.g. au10)
hostname

# 3) get jupyter notebook from anaconda or e.g. from modules:

module purge
module load Anaconda3/5.2.0

source activate lhack-an3-5.2

jupyter-notebook

# (you will get the link, e.g.: http://localhost:8888/?token=56240c.....)

# 4) open ThinLinc to aurora.lunarc.lu.se or clu0-gn-0 (MAX IV)
tlclient

# 5) in ThinLinc, open terminal and type: "ssh -L 8001:localhost:8888 au10".
# You are using info about the node (au10) and the port (8888) from the previous
# steps no. 2-3 and you choose a new local port 8001 (your choice but must be free)
ssh -L 8001:localhost:8888 au10

# 6) in ThinLink, open browser and put there a link (keep localhost ip address!!! but
# replace the port (8888) with the new one (8001)

http://localhost:8001/?token=56240c.....

# You should have the notebook running in the node (au10) but the interface in the
# browser in ThinLinc.

MAX IV modules at LUNARC

Modules with software used at MAX IV like pymca, pyFAI or NanoMAX scanViewer can be loaded on aurora as below.

# get extended MAX IV modules
module use /lunarc/nobackup/projects/snic2020-12-4/pub/sw/modules/maxiv
# load NanoMAX analysis utilities
module add foss/2018b Python/3.6.6/nmutils/0.3.0
# start NanoMAX scanViewer
scanViewer