Skip to content

Latest commit

 

History

History

v1.3

README.md

Reproducing results for Minichain v1.3

We assume that conda is installed and available in your path. To confirm, please run which conda

To reproduce the results, execute the following commands:

# Inside minichain/data/v1.3 folder
chmod +x Reproduce.py
./Reproduce.py -t8

Here -t8 represents the thread count. The memory required is about 15 GB per thread. We recommend setting the maximum thread count based on your hardware.

Benchmarking methodolody

Step 1: Data preparation

  • Download 61 major histocompatibility complex (MHC) sequences from the Human Pangenome Reference Consortium (HPRC). These are available on Zenodo.
  • We use the MHC sequence from CHM13 T2T assembly as the reference during graph construction. We do not use the MHC sequence from GRCh38.
  • Using 60 haplotypes, we build a pangenome graph by augmenting structural variations (SVs) with Minigraph.
  • Subsequently, we align each haplotype back to the pangenome graph with Minigraph using -cx asm flag. We use these alignments to store haplotype paths in the GFA file.

Step 2: Simulating haplotypes

  • We simulate 135 MHC haplotype sequences. Each sequence is a mosaic of the haplotype paths stored in the graph.
  • We add substitution errors with rates 0.1%, 1%, and 5% in the query sequences.

Step 3: Alignment and haplotype switches

  • The simulated MHC haplotypes are aligned to the MHC pangenome graph, and the chained haplotype paths are recorded with -b1 parameter.

Step 4: Evaluation metrics

  • Using the known ground truth (recombination events) and the chained haplotype paths, we measure Pearson correlation coefficient and F1-scores.

Step 5: Result visualization

  • The generated plots will be saved in a folder labeled "align_x," where 'x' represents the substitution rate.