QIBS 2026 - Module 6 workshop

Welcome to the QIBS 2026 module 6 labs! In our four sessions together, you will gain hands-on experience in analyzing single-cell sequencing data. The dates for each lab are

• Lab 1: Friday, March 25, 2026, 12-2pm ET
• Lab 2: Friday, April 3, 2026, 12-2pm ET
• Lab 3: Thursday, April 23, 2026, 9-11am ET
• Lab 4: Friday, April 24, 2026, 12-2pm ET

In this README, I will provide the necessary details to complete each lab, so please always pull the latest version of the repository at the beginning of our sessions!

Setup

I am assuming the local SSH config includes aliases to ssh into the Cayuga login node and compute nodes, i.e.,

Host cayuga-login
  Hostname cayuga-login1.cac.cornell.edu
  IdentityFile path/to/cayuga/ssh/key
  User CWID
  RequestTTY force
  ServerAliveInterval 60
  TCPKeepAlive yes

# Update HostName with assigned hostname after job submission
Host cayuga-compute
  HostName c0010
  ProxyJump cayuga-login
  IdentityFile path/to/cayuga/ssh/key
  User CWID

where CWID is your user id; please adjust any following commands if your setup differs.

General

Please setup this repository as follows

ssh cayuga-login
cd /athena/qibs_class/scratch/CWID/optional/additional/path
git clone https://github.com/dpeerlab/qibs2026.git

All provided commands assume that the qibs2026 repository is at the base of your scratch directory; i.e., make sure the adjust the commands if your setup differs.

Job submission

# ssh into cayuga login node
ssh cayuga-login
# start a screen session
screen -S qibs2026_module_6
# start an interactive session
srun -n1 --pty --partition=qibs_class --mem=16G --time=02:00:00 bash -i
# hostname to update HostName in ssh config
hostname

Following, you can connect to the Cayuga compute node via the IDE of your choice, preferably VS Code for our sessions.

Jupyter notebook setup

On the compute node, you can activate the conda environment that I am providing with

conda activate /athena/cayuga_0083/scratch/pfw4001/envs/qibs2026-py312

When you use the environment for the first time, you may have to run

conda activate /athena/cayuga_0083/scratch/pfw4001/envs/qibs2026-py312
python -m ipykernel install --user --name qibs2026-py312

in order to use the kernel for your Jupyter notebooks.

QIBS2026 repository details

For our purporse, the important parts of this repository are

• data/: this directory will contain the data we are using
• notebooks/: you will find the notebooks for each session here

In combination with the provided conda environment, the Python package corresponding to this repo qibs2026 provides easy access the data directory in Python from anywhere via the DATA_DIR variable

from qibs2026 import DATA_DIR

Lab instructions

Lab 1

In our first lab, we will implement a standard scRNA-seq analysis pipeline. The corresponding Jupyter notebook includes questions for which you will provide the answers to using the worksheet you have received.

• Data: copy the data for this lab into the data directory

cp -r /athena/qibs_class/scratch/pfw4001/data/lab_1 /athena/qibs_class/scratch/CWID/qibs2026/data/

• Notebook: notebooks/lab_1/sc_basics.ipynb
• Deadline for worksheet: April 2 at 12pm ET

Please upload the worksheet and notebook to the Google Drive we shared with you, using the file format worksheet_1_LASTNAME.xlsx and sc_basics_LASTNAME.ipynb.

Lab 2

In out second lab, we will discuss analysing single-cell perturbation data! The corresponding notebooks again includes questions for you to answer. As for the first lab, please upload the answer sheet and notebook following the same naming convention as before.

• Data: either use the DATA_DIR variable or copy the data for this lab into the data directory and import from there

cp -r /athena/qibs_class/scratch/pfw4001/data/lab_2 /athena/qibs_class/scratch/CWID/qibs2026/data/

• Notebook: notebooks/lab_2/sc_perturbseq.ipynb
• Deadline for worksheet: April 9 at 12pm ET

Installation

There is no need to install anything since we will use an existing conda environment. However, these are the commands I used to set it up

conda create -n qibs-py312 python=3.12 --yes && conda activate qibs-py312
pip install -e ".[dev,jupyter]"
pre-commit install

python -m ipykernel install --user --name qibs-py312 --display-name "qibs-py312"