GRMT

GRMT (Generative Reconstruction of Mutation Tree from scratch) is a tool for tumor mutation tree inference from single-cell DNA sequencing data. GRMT is able to efficiently recover the chronological order of mutations and scale well to large datasets. It exploits BayesOpt to estimate error rates in mutation data based on the Bayesian optimization algorithm.

Requirements

• Linux systems.
• CMake3.0+.
• g++.

Installation

To build binary, do as follows:

tar -zxvf GRMT.tar.gz
cd GRMT
cmake .
make

After the installation, the main programs of GRMT are generated in “bin” directory. Type following command if you want to add GRMT to system path:

make install

Usage

GRMT uses the genotype matrix derived from single-cell DNA sequencing data to infer mutation tree.

Example:

grmt -i testdata/example.txt -o testdata/example -a 0.01 -b 0.2

Input Files

1. Genotype Matrix

The mutational profile of single cells are denoted as a genotype matrix. Each row defines the mutation states of a single cell, and each column represents one mutation. Columns are separated by tabs. The genotype matrix can be binary/ternary.

a) Binary: Only presence/absence of a mutation is distinguished.

The entry at position [i,j] should be

• 0 if mutation j is not observed in cell i,
• 1 if mutation j is observed in cell i, or
• 3 if the genotype information is missing

b) Ternary: Heterozygous and homozygous mutations are distinguished.

The entry at position [i,j] should be

• 0 if mutation j is not observed in cell i,
• 1 if heterozygous mutation j is observed in cell i
• 2 if homozygous mutation j is observed in cell i
• 3 if the genotype information is missing

2. Cell names (optional)

A file listing the names of the single cells. Each row specifies the name of a single cell. If no such file is specified, the cells are numbered from 1 to N (N is the number of cells).

3. Mutation names (optional)

A file listing the names of the mutations. Each row specifies the name of a mutation. If no such file is specified, the mutations are numbered from 1 to M (M is the number of mutations).

Output File

Mutation tree

The mutation tree is written to a file in GraphViz format. The base name of the output file is provided by users.

Arguments

• -i, --input <filename> Replace <filename> with the file containing the genotype matrix.

• -o, --output <string> Replace <string> with the base name of the output file.

Optional arguments

• -c, --clabel <filename> Replace <filename> with the path to the file containing the names of the cells.

• -m, --mlabel <filename> Replace <filename> with the path to the file containing the names of the mutations.

• -l, --lambda <Double> Set <Double> to a value between 0.5 and 1. This specifies the value of the hyper-parameter $\lambda$. The default value is 0.7.

• -K, --kappa <Double> Set <Double> to a positive value. This specifies the value of the hyper-parameter $\kappa$. The default value is 1.0.

• -k, --maxl <INT> Set <INT> to the desired maximum number of times that a mutation can be lost. The default value is 0.

• -t, --threads <INT> Set <INT> to the number of threads to use. The default value is 1.

• -a, --alpha <Double> Set <Double> to estimated false positive rate of the single-cell sequencing experiment.

• -b, --beta <Double> Set <Double> to estimated false negative rate of the single-cell sequencing experiment.

• -n, --n_init <INT> Set <INT> to the desired number of initial data points to sample in the BO algorithm. The default value is 100.

• -N, --n_iter <INT> Set <INT> to the desired number of data points to evaluate in the BO algorithm. The default value is 30.

Contact

If you have any questions, please contact [email protected].