A Novel Haplotype Reconstruction Method (MLHapRec)

MLHapRec is a maximum likelihood approach for haplotype reconstruction which uses the SAMFIRE multi-locus variant calling package for processing short-read data where samples are collected during a transmission of a virus (like influenza) from a donor and a recipient population (i.e. two time points).

This code is originally developed for estimating the transmission bottleneck size of flu, but can be used for any other set of short-read data for the purpose of constructing the underlying (unkonw) haplotypes and estimating bottleneck size.

Requirements

This code requires a Multi_locus_trajectories.out file and an inferred noise parameter C from SAMFIRE. An initial seed value and number of attempts for reconstruction can also be provided. In case you do not wish to provide a seed or number of attempts, put 0 as the input for both and a default seed = 1 and attempts = 20 will be used to run the code. For every time a new haplotype is added to the list, number of attempts will determine how many times a set of candidate haplotypes are optimised before the code accepts the list and compares the BIC value with the previous step.

Usage

Given that your Multi_locus_trajectories.out is in the /path/to/directory/, your inferred noise parameter is, say, C = 660, seed = 123, and attempts = 10, you can run the code by typing:

./run_MLHapRec /path/to/directory/Multi_locus_trajectories.out 660 123 10

If, say, you do not wish to provide a seed number, you can run the same code by typing:

./run_MLHapRec /path/to/directory/Multi_locus_trajectories.out 660 0 10

and a default seed value = 1 will be used.

Output

Running the code, typically, takes anywhere from 10sec to 10h depending on the size of Multi_locus_trajectories.out and number of attempts taken. For instance, if there are about 10 partial haplotype reads in the Multi_locus_trajectories.out and 20 attempts are made for reconstruction each time, it takes up to ~15min for the full haplotype reconstruction.

After a successful execution, MLHapRec generates the following three files:

• update.txt shows how the optimisation process takes place at every round as a new haplotype is added to the list of inferred haplotypes
• raw_haplotypes.txt lists all the reconstructed haplotypes with their frequencies before and after transmission.
• outcome_1.txt lists the haplotypes from raw_haplotypes.txt which have not acquired a mutation once established in the recipient -- a mutation is defined to be a haplotype with frequency <1e-7 in the donor and >1e-7 in the recipient. Note that in the case of no mutation, raw_haplotypes.txt and outcome_1.txt are essentially identical with a slight difference in that in outcome_1.txt, we set all the frequencies below 1e-7 in raw_haplotypes.txt to zero.