Instructions for using the SNP-Slice package

This repository contains the implementation of the SNP-Slice algorithm, which is a Bayesian nonparametric method to resolve multi-strain infections. You can find the motivation for this problem, a description of the algorithm, as well as our results in the Bioarxiv preprint titled SNP-Slice Resolves Mixed Infections: Simultaneously Unveiling Strain Haplotypes and Linking Them to Hosts (https://www.biorxiv.org/content/10.1101/2023.07.29.551098v2).

Preparing your local directory to run SNP-Slice.

The structure of the directory contains:

• snpslicemain.R (the main execution file).
• inputdata/ (a directory to store input data files, named prefix_read1.txt, prefix_read0.txt and prefix_cat.txt.
• output/ (a directory to store output data (A,D))
• mcmcRData/ (a directory to store RData files for warm start)
• source/ (a directory containing the actual implementation of the algorithm).

Using the algorithmn.

1. First of all, specify a prefix in snpslicemain.R.

   For example, setting prefix <- "scenario1" on line 21 of snpslicemain.R, the script will read scenario1_read1.txt and scenario1_read0.txt from the inputata directory.

2. Now you can run the algorithm in the command line, with, for example,

Rscript snpslicemain.R model=3 nmcmc=10000 alpha=2 gap=100.

3. You can also decide which model to use, by controlling the value of model. We recommend setting model in the command line instead of in the execution file. The default value is Negative Binomial model. This is the codebook:

• model <- 0 for the cat model
• model <- 1 for the Poisson model
• model <- 2 for the Binomial model
• model <- 3 for the Negative Binomial model.