PhylogicNDT container contains complete build for Dockerfile of PhylogicNDT from Broad Institute that can be easily used to analyse tumour evolution. It has been built to automatically run Cluster and BuildTree modules, which required to determine the phylogenetic relationship between tumour subclones.
PhylogicNDT image can be found on aalhendi1707/phylogicndt, and it is only for non-commersial use. To pull the latest version of it, please run this command on terminal:
docker pull aalhendi1707/phylogicndt:latest
docker run -it --rm \
-e MODULE="Cluster" \
-e PATIENT_ID="PatientXX" \
-e N_ITER=1000 \
-e MIN_COVERAGE=30 \
-e CALC_CCF="YES" \
-e IMPUTE="NO" \
-e BUILDTREE_CCF_THRESHOLD=0.1 \
-e BLOCKLIST_CLUSTER="None" \
-v /path/to/data:/INPUTDIR \
aalhendi1707/phylogicndt:latest
-e MODULE to select running module. This version of docker container only support Cluster and BuildTree. Default used by algorithm Cluster.
-e PATIENT_ID to set Patient/Case ID.
-e N_ITER" Number iterations that will be used in Cluster and BuildTree modules. Default used by algorithm is 1000.
-e MIN_COVERAGE Mutations with coverage lower than this will not be used to cluster and instead re-assigned after dp clustering.
-e CALC_CCF Flag must be set to calc_ccf and sample purity must be provided. Also local copy number must be attached to each mutation in the maf with columns named local_cn_a1 and local_cn_a2 represent minor and major allele fractions respectively.
-e IMPUTE Assume 0 ccf for missing mutations.
-e BUILDTREE_CCF_THRESHOLD ccf threshold for blacklisting clusters for a BuildTree and Cell Population. Default used by algorithm is 0.1.
-e BLOCKLIST_CLUSTER List cluster ids to blacklist from BuildTree and CellPopulation. Default used by algorithm None.
-v /path/to/data The path to your Input/output directory. This directory must contains (1) Somatic variants in maf file format for each sample [required], (2) MySimulation_input.sif [required] as described in below, (3) Driver_genes_v1.0.txt list for dirver genes that would be used to annotate for driver mutations [optional], as PhylogicNDT already comes with list of driver genes. The outputs will be written to this directory as well.
When running the Cluster module, prefer to use the sif file option. The command line code is much cleaner, and the format provides a much more organized way to keep track of your input files and values.
Sif file has 4 required columns, and 1 optional column:
- sample_id: the ID of each sample from the patient.
- maf_fn: a truncated version of the MAF file with/without additional CCF information.
- seg_fn (optional): a truncated version of allelic seg calls.
- purity: the purity of each tumor sample from the patient.
- timepoint: if temporal ordered by biopsy time, if spatial then arbitrary
Below is the accepted format for MAF file
| Hugo_Symbol | Chromosome | Start_position | Reference_Allele | Tumor_Seq_Allele2 | t_ref_count | t_alt_count | Protein_change | Variant_Classification | Variant_Type | local_cn_a1 | local_cn_a2 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| CASZ1 | 1 | 10705011 | G | C | 200 | 13 | p.F1277L | Missense_Mutation | SNP | 0 | 1.4827313901312 |
| CAMK2N1 | 1 | 20811852 | G | C | 66 | 31 | p.Y7X | Nonsense_Mutation | SNP | 0 | 1.4827313901312 |
| RPL11 | 1 | 24020360 | T | C | 250 | 136 | p.V74A | Missense_Mutation | SNP | 0 | 1.4827313901312 |
| CDKN2C | 1 | 51436117 | A | T | 53 | 230 | p.Q26L | Missense_Mutation | SNP | 0 | 1.4827313901312 |
The BuildTree inputs are just a subset of the outputs from the Cluster module: Again, we need to proide the patient ID and the sif file. A description of the mutation CCF file and cluster CFF file output by the cluster module are described below.
- mutation_ccf_file: This file contains the cluster assignment of each mutation in the MAF in the maf_fn file.
- cluster_ccf_file: This file contains CCF information on each cluster, such as the average CCF of mutations in the cluster and the probability distribution of the cluster CCF.