Quantum Computing for Multi-omics Analyses (ISMB 2026)#

../_images/ISMB26banner.png

ISMB 2026 Tutorial — Washington, DC (July 12, 2026)

An in-depth, hands-on tutorial exploring how quantum computing enables advanced multi-omics analysis and hybrid ML workflows.

Overview#

Join us for an interactive full-day tutorial covering the foundations and applications of quantum computing (QC) in multi-omics data analysis.

Participants will:

  • Learn how to preprocess and encode biological data for quantum algorithms

  • Explore quantum machine learning (QML) and hybrid models

  • Understand data complexity measures to assess when QC can outperform classical approaches

  • Work hands-on with real-world datasets and QBioCode

Instructors#

Aritra Bose — Staff Research Scientist, IBM Profile

Filippo Utro — Senior Research Scientist, IBM Profile

Laxmi Parida — IBM Fellow Profile

Learning Objectives#

Participants will:

  • Understand quantum computing fundamentals (states, circuits, gates)

  • Learn preprocessing of multi-omics data for QML

  • Analyze data complexity and ML limitations

  • Apply quantum ML and hybrid pipelines

  • Benchmark quantum vs classical models

Prerequisites#

Set up the environment (do this before the workshop)

Clone the repository and install QBioCode in editable mode. Installing with pip install -e . reads requirements.txt and makes the qbiocode package importable inside the notebooks — a plain pip install -r requirements.txt does not install the package itself, so the import qbiocode cells would fail.

git clone https://github.com/IBM/QBioCode.git
cd QBioCode

# create an isolated environment (venv or conda), e.g.
python -m venv .venv && source .venv/bin/activate

# install QBioCode + all runtime dependencies
pip install -e .

Requires Python 3.10–3.12. This installs everything the hands-on notebooks need, including the single-cell preprocessing stack (scanpy, anndata, leidenalg, igraph). The datasets used in the hands-on sessions ship with the repository, so no extra downloads are required.

Agenda#

Format

Expand each item to access description, slides, and hands-on material.

Session I — Foundations (09:00–10:45)#

1. Introduction (09:00–09:15)

Overview of tutorial goals, structure, and expected outcomes.

Materials

2. Quantum computing fundamentals with Qiskit (09:15–09:45)

Introduction to: - quantum states, gates, circuits - basic Qiskit workflows

Includes short hands-on demo.

Materials

3. Data complexity measures (09:45–10:15)

Understanding intrinsic dataset complexity and how it impacts learning.

Materials

4. QBioCode application setup (10:15–10:45) — 🖐️ Hands-on

Install and configure the QBioCode environment.

What QBioCode is

QBioCode is a modular toolkit for benchmarking classical, quantum, and hybrid ML on multi-omics data. The modules you’ll touch today:

  • QProfiler — profiles a dataset’s complexity (Fisher ratio, mutual information, intrinsic dimension, …) and benchmarks classical models against quantum methods (e.g. the projected quantum kernel, PQK, QNN, etc.).

  • QSage — meta-learning for automated model selection.

  • QEnsemble — combines classical and quantum models.

Materials

☕ Coffee Break (10:45–11:00)

Session II — QBioCode Applications (11:00–13:00)#

1. Qprofiler in multi-omics data (11:00–11:45) — 🖐️ Hands-on

Hands-on session using QProfiler for biological datasets.

Materials

2. QBioCode Module I: Quantum Projection Learning (11:45–12:15) — 🖐️ Hands-on

Apply Projected Quantum Kernels (PQK) followed by classical ML for multi-omics datasets.

Materials

3. QBioCode Module II: Quantum ensemble methods (12:15–13:00) — 🖐️ Hands-on

Combining classical and quantum machine learning models via ensemble methods for improved performance.

Materials

🍽️ Lunch Break (13:00–14:00)

Session III — Quantum Algorithms & Applications (14:00–16:00)#

1. Quantum Algorithms for Healthcare and Life Sciences (14:00–14:40)

Quantum Algorithms for Biomedical and Translational Applications.

Materials - 📄 Slides: Quantum Applications

2. Team formation & task selection (14:40–14:45) — 🖐️ Hands-on

Choose 1–2 tasks from the task pool, or propose your own.

3. Quantum Subway Mapping (14:45–15:15) — 🖐️ Hands-on

Quantum Subway hands-on activity.

Materials - 📄 Slides: Quantum Subway

4. First readouts (15:15–15:30) — 🖐️ Hands-on

Teams read out — maximum of 2 minutes each.

5. Dataset introduction (15:30–15:45)

Presentation of the dataset used in the hands-on session: the 10x Genomics

5k PBMC CITE-seq reference, and how it is QC’d and turned into the benchmark tasks.

Dataset — 5k PBMCs from a healthy donor with cell-surface proteins

  • Source: 10x Genomics — 5k PBMCs (Next GEM 3′ v3.1, TotalSeq-B)

  • Sample: peripheral blood mononuclear cells (PBMCs) from a single healthy human donor

  • Assay: Chromium Next GEM Single Cell 3′ v3.1 with Feature Barcoding (CITE-seq) — paired Gene Expression (RNA) and Antibody Capture (ADT) measurements

  • Antibody panel: 31–32 TotalSeq-B surface-protein markers (e.g. CD3, CD4, CD8a, CD14, CD16, CD19, CD56, HLA-DR)

  • Scale: ~5,527 cells detected · 33,538 genes + 32 ADTs · ~30,853 reads/cell (Illumina NovaSeq); processed with Cell Ranger 3.1.0

  • License: CC BY 4.0

QC & task construction

Starting from the 10x filtered_feature_bc_matrix.h5 (Gene Expression + Antibody Capture), the pipeline runs standard scRNA QC (mito/ribo/hb metrics, cell/gene filtering, Scrublet doublet removal), normalizes RNA and protein, computes RNA PCA and protein embeddings, and builds two KNN view-graphs (RNA on the 8-D PCA space, protein on the full 32-marker scaled space) for the quantum-walk analysis. Leiden clustering + marker/ADT annotation then yields three leakage-aware

binary task — cd4_vs_cd8 — plus small balanced .h5ad benchmark file.

Materials

6. Run Quantum and Classical Machine Learning methods (15:30–16:00) — 🖐️ Hands-on

Open the notebook for the module your team chose and run it top-to-bottom on the CD4 vs CD8 classification task. Set your options in the Configuration cell, then run. Both notebooks run on the local statevector simulator and finish on a laptop in a few minutes.

Materials

☕ Coffee Break (16:00–16:15)

Session IV — Hands-on + Discussion (16:15–18:00)#

1. Continue implementation & results analysis (16:15–16:45) — 🖐️ Hands-on

Extend experiments and interpret results.

Materials - 💻 Notebook: QProfiler — single-cell binary (CD4 vs CD8)

2. Long Readout (16:45–17:15) — 🖐️ Hands-on

Each team reads out results and comparison with the classical baseline, and assesses what was learned.

2. Interactive Q&A (17:15–17:45)

Discussion, feedback, and open questions.

3. Concluding remarks (17:45–18:00)

Key takeaways and future directions.

Materials

Materials#

Audience#

  • Computational biologists

  • Bioinformaticians

  • Data scientists in life sciences

  • Clinicians and practitioners

Suitable for early- to senior-level researchers