/celldive

Primary LanguageJupyter NotebookMozilla Public License 2.0MPL-2.0

CellDive Gene-level analysis of lymphocyte single-cell RNA data based on a clonality specification language

CellDive (src/) is a tool for exploring clonality of lymphocytes and differential expression based on single-cell RNA-seq:

  • clonality rules specification based on the alpha and beta chains of T-cell and B-cell receptors (TCR and BCR)
  • clustering into main-clone, related-to-main-clone, bystander groups and single bystanders based on the clonal rules
  • bipartite graph visualization of the full clonality information from the samples
  • differential expression (DE) between specified clones within the same sample (no batch effects!)
  • gene sets overview accross samples and DE methods

CellDive has been applied (example/) to a dozen of human cutaneous lymphoma samples. The clonality rules were used to partition each sample into a malignant clone (main-clone) and healthy cells (bystanders). Differential expression analyses was applied to the two clusters (no batch effect).

Workflow

License: MPL 2.0

Contributions

Pesho Ivanov (advisor Martin Vechev), Software Reliability Lab, ETH Zurich:

  • code and execution
  • alpha and beta chain language rules for specifying clones. each cell is colored depending on the clone
  • bipartite graph with left nodes for alpha-chains, right nodes for beta-chains, and cells as edges (possibly connected to No-alpha and No-beta nodes)

Bipartite graph

Data used to produce example/, Department of Dermatology, University Hospital Zurich:

  • Yun-Tsan, Desislava Ignatova, Emmanuella Guenova

Note: The code has been written in 2017--2018 so it may be outdated.