Note: This repository contains scripts that were developed for "SCsnvcna: Integrating SNVs and CNAs on a phylogenetic tree from single-cell DNA sequencing data".

Authors: Liting Zhang, Hank Bass, Jerome Irianto, and Xian Mallory.

SCsnvcna is an algorithm that integrate SNVs and CNAs on a phylogenetic tree from single-cell DNA sequencing data. SCsnvcna uses a loss-supported model that allow mutation lossess based on observed copy number loss on CNAs tree.

SCsnvcna is implemented in Python.

1. Setup

2. Running SCsnvcna

   • Input
   • Output
   • Usage

3. Simulation

To use SCsnvcna, first clone the SCsnvcna repository locally.

git clone [email protected]:compbio-mallory/SCsnvcna.git

SCsnvcna uses Python and doesn't require any compilation.

SCsnvcna takes as input three files. The first describe the copy number tree structure from CNAs cell. The second descibes the observed genotype matrix from SNV cells. The third describes the set of supported losses for each edge in the copy number tree.

1. Tree structure file. This file describe the copy-number tree structure constructured from CNAs cells. This file takes the format of a tab-separated edge list, where the first two columns are child and parent nodes, followed by edge ID connecting into child node (same as child node), edge length, percentage of CNAs cells below this node, whether it is a leaf, and an optional node name. If node names are given, node names will be used to generate the output newick file instead of node ids.

 0	-1	0	0.001	1	0	[in0]
 1	0	1	0.002	0.6	0	[in1]
 2	0	2	0.04	0.4	1	[cell0]
 3	1	3	0.05	0.3	1	[cell1]
 4	1	4	0.4	0.3	1	[cell2]

2. Genotyp matrix file. This file describe the observed genotype matrix from SNVs cells. This file takes the format of a tab-seperated matrix. Where each line is the genotype for each cell, whereas each columns are genotype for each SNVs. 0 is absent, 1 is present, and 3 is missing.

0	1	3	0	...
1	1	1	0	...
...

1. Mutation loss support file. This file describes the potential mutation loss of each edge id. The first column is edge, the second column is the SNV that might be lost on this edge.

1	19
12	55
...

2. Reveal file. This file describes which CNA nodes should a SNV cell be constrained to. The first column is edge id, the second columns is the SNV cell id.

24	10
26	10
30	11
...

3. SNV cell names file. This file includes the SNV cell ID and its corresponding SNV cell name separated by tab. If this such a file is provided, SNV cells names will be used when generating newick file instead of ids.

0	MA27
1	MA28
...

4. SNV names file. This file includes the SNV ID and its corresponding SNV name separated by tab. If this such a file is provided, SNV names will be used when generating newick file instead of ids.

0	TP53
1	NRAS
...

Examples of these files can be found in the data/simulation/ directory.

Scsnvcna produces several output files, as detailed below. In addition to these, SCsnvcna produces several auxiliary files that are used for graphing purposes and log file are not detailed below.

1. Binary mutation matrix file ([prefix].G). This file describes the binary presence (1) or absence (0) of each mutation in each cell. This is a tab-separated file where rows correspond to cells, of the following format.

   [mut1] [mut2] ... [mut_m]
   ...
   

2. SNV placement file (prefix.snv). This file describe the placement of SNVs. The first column is the SNV ID, and the second column is where the SNV is placed on.

   SNV1	Edge1
   ...
   

3. SNV cell placement file (prefix.cell). This file describe the placement of SNVs cell. The first column is the SNV cell id, and the second column is where the SNV cell is placed on.

   cell1	leave1
   ...
   

4. Log probability of the selected tree (prefix.Prob). This file describe probability of the selected tree.

   Probability
   

SCsnvcna can be run from the command line as follows.

python3 code/main.py -tree data/sample_data/tree_8_p3.tsv -D data/sample_data/input.D.tsv -overlap data/sample_data/input.mutCNAoverlap.tsv -out data/sample_data/prefix -alpha 0.01 -beta 0.2 -sigma 0.05 -itr 100000 -reveal data/sample_data/input.SNVcell.tsv -restart 5

• -tree Copy number tree structure file. Required.
• -D observed genotype matrix for SNV cells. Required.
• -out output prefix. Required.
• -overlap files contain potential mutation loss. Optional.
• -alpha initial false positve rate. Default is 0.01. Optional
• -beta initial false negative rate. Default is 0.2. Optional
• -sigma standard deviation of CPs between CNA cells and SNV cells. Default is 0.05. Optional.
• -itr number of iteration. Default is 100000. Optional.
• -reveal file contains where should SNV cells be constrained to. Optional.
• -restart number of restart. Default is 5. Optional.
• -burnin number of iterations to burn in. Default is 2000. Optional.
• -SNVcell file contains the SNV cells IDs and its corresponding names. Optional.
• -SNVID files contains the SNV IDs and its corresponding names. Optional.

Rscript code/plot_SCsnvcna.R prefix.newick prefix.jpeg

• prefix.treeText are the newick output of SCsnvcna.
• prefix.jpeg output file name for the plot

To tune the plot, change the parameter at the end of the plot_tree.R accordingly.

1. Generate a Tree

python3 code/simulation_code/gen_tree.py -F 8 -B 0.3 -o tree_8_p3.tsv

• -F width of the tree. Number of leaves on this tree. Default is 16.
• -B Beta splitting variable. Default is 0.3.
• -o Output file.

2. Add mutations and cells to the tree.

python3 code/simulation_code/sim_par.py -c 50 -n 20 -f tree_8_p3.csv -P input

• -c Number of cells. Default is 100.
• -n Number of mutation. Default is 100.
• -f Tree file from step 1.
• -P Output file prefix.