A Complete Demo


Here’s a demo introducing a complete ELLA analysis pipeline.

The script that will be used in this demo should have already been downloaded (while cloning the ELLA repo). The data (complete_demo_data.pkl) that will be used can be downloaded from here. You should be able to orgnize these at your local ELLA folder as follows:

ELLA/scripts/demo/complete_demo/
├── lightning_logs
│   └── run1
├── log
├── complete_demo_data.pkl
├── complete_demo_postprocess.ipynb
├── prepared_data
└── run_complete_demo.sh

The data is a subset of the processed seqFISH+ embryonic fibroblast dataset. The input data (complete_demo_data.pkl) mainly contains a dictionary of three dataframes corresponding to gene expression, cell segmentation, and nucleus segmentation (optional) with 20 cells and 50 genes.

ELLA Anlysis

  1. Preprocess 
    python -m ella.data.prepare_data -i your_dir/ELLA/scripts/demo/complete_demo/complete_demo_data.pkl -o your_dir/ELLA/scripts/demo/complete_demo/prepared_data
  2. Run ELLA 
    bash run_complete_demo.sh

    The corresponding recipe is complete_demo.yaml.

  3. Postprocess
    Using complete_demo_postprocess.ipynb.
    Specifically, we can now cluster the estimated (significant) expression intensities into clusters of patterns. We find the optimal number of kmeans clusters K with the ELBOW method where K is chosen as a point where the distortion/inertia begins to decrease more slowly.

Based on the plots, it seems 5 can be a proper choice, thus let’s proceed with K=5 to obtain 5 pattern clusters:

Pattern 1: 12 genes
Pattern 2: 7 genes
Pattern 3: 11 genes
Pattern 4: 9 genes
Pattern 5: 6 genes

Plots: numbers and proportions of significant genes, estimated expression patterns, and estimated pattern scores

We can overlay all genes in the same cluster in cells to have a more intuitive sense of the patterns.