The goal is to generate posterior samples of transmission trees in order to estimate epidemiological parameters of interest.

WARNING --- the current implementation is heavily optimized for running SARS-CoV-2 analyses.

To install this package, you will need remotes packages:

install.packages("remotes")

Once you have successfully install the remotes package, you can run:

remotes::install_github("andersgs/transtreesampler")

You can optionally specify the Ncpus option to speed things up a bit:

remotes::install_github("andersgs/transtreesampler", Ncpus=8)

The easiest way to run an analysis with this package is to use NextFlow. To install NextFlow, I recommend you use a package manager like Brew or Conda:

brew install nextflow

conda install -c bioconda nextflow

Once you have the environment with R and NextFlow installed, as well as the package, you can get do a test run with the following command (note: this assumes that you have clusters of samples with at least 15 cases, the default minimum number of cases for a cluster to be analyzed --- see below how to change this). This will run a very short MCMC chain (50 iterations sampled every 1 iteration, this is not sufficient for any analyses, just to test that things are working):

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa --metadata /full/path/to/metadata.csv

The metadata file must have the following columns and data, and must be in CSV format (this will change in subsequent versions to be more generic --- but see below if your column headers don't exactly match the headers in the table):

The multiFASTA file assumes the sequences are aligned, and have identifiers that match the identifiers in the vic_id column of the metadata file:

>seq1
ATCG
>seq2
ATCC
>seq3
TTCC
>seq4
TTGC

That is ok, you can change that by using the following options (change only those that apply):

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa \
                     --metadata /full/path/to/metadata.csv \
                     --cluster_col <my_cluster_id> \
                     --sample_id_col <my_sample_id> \
                     --sampling_date_col <my_sample_date_col> \
                     --onset_date_col <my_date_onset_col>

To change the minimum size of clusters to be analyzed (default 15), you need to specify the --min_cluster_size flag:

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa \
                     --metadata /full/path/to/metadata.csv \
                     --min_cluster_size 20

If you have metadata with several clusters, but you only want to run a single one of them through tree sampling process, you can use the --filter_cluster_id option to select the cluster of interest. In the code below, we restrict the analysis only to the cluster identified as 67:

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa \
                     --metadata /full/path/to/metadata.csv \
                     --filter_cluster_id 67

Let say you want to run some quick tests, and don't want to run a full set of samples in your clusters. To restrict the number of samples included in the analysis, you can use the --max_samples option. Below, we restrict our analysis to the first 6 samples in each cluster:

NOTE: This option really should only be used for testing.

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa \
                     --metadata /full/path/to/metadata.csv \
                     --max_samples 6

To run a proper length MCMC, you need to specify two options that set the length of the MCMC and the thinning (how frequently samples are kept):

nextflow run andersgs/transtreesampler --fasta /full/path/to/fasta.fa \
                     --metadata /full/path/to/metadata.csv \
                     --n_iter 100000 \
                     --sample_every 100

The outputs of each step of the workflow will be stored in a results folder within your project directory. Each step of the workflow will store its outputs in a sub-folder of the results folder named after that sub-step:

• data sub-folder will store the output of the first step where the metadata and sequence data are merged, and clusters of size greater than or equal to the minimum size are identified. This first step will output as many files as there are clusters that meet the minimum size criterion.

• likelihood sub-folder will store the output calculating occurrence of all the possible pairwise base permutation patterns (e.g., AA, CC, TT, GG, AC, etc.) --- this tally is used in the likelihood function and calculating it once avoids having to calculate repeatedly throughout the MCMC. The calculation is added to the data from the previous step.

• conditions sub-folder adds to the data from the previous step different assumptions about the w_dens and f_dens distribution. At the moment, four conditions are tested (all geared towards SARS-CoV-2). This leads to four files per cluster.

• mcmc sub-folder stores the output for each MCMC run.

• tidy_output sub-folder stores the output of the MCMC data in a tidy format. Note that here, all column names produced by outbreaker2 are prefixed with outb2_, and all column names originated from the data are prefixed with meta_

• cluster_output sub-folder groups data from different conditions from the same cluster into a single data.frame, and outputs it as a CSV file. Here, some column names are changed to match requirements for downstream analyses.

• Kurnia Susvitasari
• Jessica Stockdale
• Ben Sobkowiak
• Anders Goncalves da Silva
• Paul Tupper
• Caroline Colijn