jasperyh / lineage_tracing

Allen Institute Cell Lineage Reconstruction DREAM Challenge

A Lineage reconstruction method for DREAM challenge

This is team Jasper06's submission to Allen Institute Cell Lineage Reconstruction DREAM Challenge

Method description

We used probability-based hierarchical clustering to reconstruct the cell lineage tree. In short, the transition rate of the initial state to altered states are derived using the ground truth. Next, we derived the probability of cells arising from different parents using transition rates. Then, the pairwise cell distance is calculated using the probability. Finally, we used hierarchical clustering (UPGMA) to reconstruct the cell lineage.

Usage

Download data

This repository only provides 1 example dataset from each subchallenge. Download the complete dataset from here.

Running this lineage reconstruction algorithm takes the exact 4 steps for each subchallenge.

Step 1:

Use Jupyter notebook to run 01_mutation_summary.ipynb

Step 2: Derive mutation rate

Rscript 02_mutation_rate.R

Step 3: Use probability-based hierarchical clustering to reconstruct lineage tree

Use Jupyter notebook to run 03_probability_HC.ipynb

Step 4: Evaluate reconstruction result by calculating the RF score and Triplet score

Rscript 04_evaluation.R