ACDC documentation

Welcome to the ACDC documentation. This includes API references, tutorials, and guides.

acdc_py 🤘

Automated Community Detection of Cell Populations in Python

This repo contains the current Python implementation of ACDC, an optimization-based framework to automatize clustering of cell populations from scRNA-seq data using community detection algorithms. acdc_py is currently under development and new functionalities will be released, following completion and benchmarking. acdc_py is deployed as a Python package and fully compatible with Scanpy.

• Several graph-based clustering algorithms are available within acdc_py, including Leiden and Louvain.

• 2 optimization routines for parameter tuning are available, Grid Search and(generalized) Simulated Annealing.

• Optimization variables include the number of nearest neighbors, k, resolution, res, and the number of principal components, PCs.

• Several objective functions are available, including the Silhouette Score (default).

New releases will expand functionalities to new features, including the possibility to iteratively sub-cluster cell populations to find fine grain and biologically meaningful clustering solutions.

To receive updates when novel functionalities are released, feel free to add your email to the following form: https://forms.gle/NCRPJPmXzfbrMH7U7

STAY TUNED FOR UPDATES AND NOVEL DEVELOPMENTS!🤘🏾

**Please, be aware that while this project is “work in progress” and outcomes are continuously benchmarked, cross-platform compability might not yet be guaranteed.

Installation

pypi

pip install acdc-py

local

git clone https://github.com/califano-lab/acdc_py/
cd acdc_py
pip install -e .

… Start playing around! 🎸

References

1. Kiselev, VY, Andrews, TS, Hemberg, M. (2019) Challenges in unsupervised clustering of single-cell RNA-seq data. Nat Rev Genet 20, 273–282.

2. Blondel, V D, Guillaume, J, Lambiotte, R, Lefebvre, E (2008). Fast unfolding of communities in large networks”. Journal of Statistical Mechanics: Theory and Experiment. (10) P10008.

3. Satija R, Farrell JA, Gennert D, Schier AF, Regev A (2015). “Spatial reconstruction of single-cell gene expression data.” Nature Biotechnology, 33, 495-502.

4. Traag, V.A., Waltman, L. & van Eck, N.J. (2019) From Louvain to Leiden: guaranteeing well-connected communities. Sci Rep 9, 5233.

5. Xiang, Y., Gubian, S., Suomela, B.P., & Hoeng, J. (2013). Generalized Simulated Annealing for Global Optimization: The GenSA Package. R J., 5, 13.

Contacts

Alexander Wang - aw3436@cumc.columbia.edu

Luca Zanella - lz2841@cumc.columbia.edu

Alessandro Vasciaveo - av2729@cumc.columbia.edu

Contents:

• ACDC in Python
• acdc_py
• acdc_py.GS
• acdc_py.SA
• acdc_py.get_opt
  • SA_clustering()
  • SA_params()
  • SA_metric_value()
  • SA_metric_search_data()
  • GS_clustering()
  • GS_params()
  • GS_metric_value()
  • GS_metric_search_data()
• acdc_py.pp
  • corr_distance()
  • neighbors_knn()
  • neighbors_graph()
• acdc_py.pl
  • GS_search_space()
  • SA_search_space()
  • metric_vs_n_clusts()
  • silhouette_scores()
• acdc_py.tl
  • cluster_final()
  • extract()
  • merge()
  • rename()
• acdc_py.config
  • set_SS_bootstraps()
  • set_SS_weights()
  • set_clust_alg()
  • set_corr_distance_dtype()

Tutorials:

• Clustering human single-cell RNA seq data with ACDC
  • Steps 1-2: Install ACDC and its dependencies
  • Step 3. Import necessary Python packages
  • Step 4: Load the expression matrix and regulatory network using Pandas and pyVIPER
  • Step 5: Preprocessing Data
  • Step 6. Compute VIPER (Virtual Inference of Proteins by Enriched Regulon analysis)
  • Step 7. Identify clusters of cells with ACDC
  • Step 8. Compare clusters vs SS
  • Step 9. Compute best solution with 4 clusters
  • Step 10. Classify the major lineages of the clusters.
  • Step 11. Identify remaining subclusters
  • Step 12. Compute the differentiation velocity

Indices and tables

• Index

• Module Index

• Search Page