WIKI single‐cell‐DNA

Welcome to the single-cell-DNA wiki!

Pipeline Goal:

Perform single-cell DNA-seq analysis from FastQ files to figures file for missionbio tapestri data.

Steps available:

• Alignment
• Preprocessing (filtering bad quality variants, CNV and cells)
• SNV_CNV (Normalization dimension Reduction and clustering)
• PROTEIN (Normalization dimension Reduction and clustering)
• ALL (Combining DNA-seq analysis & Proteomic analysis)
• Phylogeny (reconstruction of mutations events)

Usage

---

Usage on Flamingo, the GR's computing cluster

❗ if you already used the single-cell RNA-seq pipeline it is identical

• make the parameters file according to your needs (see below how to configure the parameter file)
• indicate the path to this file in the path_to_configfile variable
• run the snakemake command

module load singularity

path_to_configfile="<path/to/your_configfile.yaml>"
path_to_pipeline="<path/to/single-cell-dna-seq>"

snakemake --profile ${path_to_pipeline}/profiles/local -s ${path_to_pipeline}/Snakefile --configfile ${path_to_configfile}

Configuration

---

1. steps & alignment: choose the steps to run

name	description	example	default value	possible value
steps	steps to run	[Aligment,preprocessing,SNV_CNV,ALL,phylogeny]	NA	Aligment,preprocessing,SNV_CNV,PROTEIN,ALL,phylogeny
tmp	temporary directory	/tmp	NA	NA
sample	sample(s) to run	[sample_1,sample_2]	NA	NA
reference_genome_path	path of the reference genome	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/hg19/ucsc_hg19.fa"	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/hg19/ucsc_hg19.fa"	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/hg19/ucsc_hg19.fa","/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v3/hg19/ucsc_hg19.fa"
reference_genome	reference genome release	"hg19"	"hg19"	"hg19"
type_analysis	select your analysis dna or dna+protein"	"dna+protein"	NA	"dna","dna+protein"
panel_path	path of your panel of variants	"</your/path/to/panel/file/location>"	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/panels/Myeloid"	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/panels/Myeloid","<your/path/panel/location>"
panel_protein_path	path of the reference fasta for protein	"/mnt//beegfs/pipelines/single-cell_dna/tapestri_database/v2/panels/protein"	"/mnt//beegfs/pipelines/single-cell_dna/tapestri_database/v2/panels/protein"	"/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v2/panels/protein/","/mnt/beegfs/pipelines/single-cell_dna/tapestri_database/v3/panels/protein/"
design_file	path to design file in order to create proper yaml file for aligment	"/your/path/to/panel/file/location"	NA	NA

2.filtering: filtering - remove bad quality variants & preprocess your data

name	description	example	default value	possible value
filter_na	filtering Missing Value	True	False	True/False
filter_na_percent	remove variants which missing value are superior or equal to the threshold	35	25	any integer
predict_missing_value	KNN predict missing value variants	True	False	True/false
filtering_variants	multiple filter in order to remove bad quality variants & cells	NA	NA	NA
max_vaf_percent	filter variants which mean VAF value is superior or equal	95	95	any integer
whitelist	variants that must be keep (even if their quality is poor)	["chr20:33868702:T/C"]	NA	NA

2.1. filtering: filtering_variants

name	description	example	default value	possible value
min_dp	The minimum depth (DP) for the call to be considered	10	10	any integer
min_gq	The minimum genotype quality (GQ) for the call to be considered	30	30	any integer
vaf_ref	All reference calls (NGT = 0) with VAF > vaf_ref are converted to no calls (NGT = 3)	5	5	any integer
vaf_het	All hetrozygous calls (NGT = 1) with VAF < vaf_het are converted to no calls (NGT = 3)	35	35	any integer
vaf_hom	All homozygous calls (NGT = 2) with VAF < vaf_hom are converted to no calls (NGT = 3)	95	95	any integer
min_mut_prct_cells	The minimum percent of the total cells in which the variant should be mutated,	1	1	any integer
min_prct_cells	The minimum percent of total cells in which the variant should be present	50	50	any integer

3.SNV: snv_norm_dimred

name	description	example	default value	possible value
method_dimred	select dimension reduction for variants matrix between	pca	pca	fa,pca
max_dims	maximum dimensions for the dimension reduction	6	6	any integer
clustering_method	clustering method to use	leiden	dbscan	graph-community,leiden,dbscan,hdbscan

3.CNV: cnv_norm_dimred

name	description	example	default value	possible value
max_dims	maximum dimensions for the dimension reduction	6	6	any integer
clustering_method	clustering method to use	leiden	dbscan	graph-community,leiden,dbscan,hdbscan

4.PROTEIN: prot_norm_dimred

name	description	example	default value	possible value
normalization	normalization method to correct noise	DSB	CLR	CLR,DSB,asinh,NSP
clustering_method	clustering method to use	leiden	dbscan	graph-community,leiden,dbscan,hdbscan

5.ALL: all_norm_dimred

name	description	example	default value	possible value
snv	SNV parameters to keep in order to combine multi-omics data	NA	NA	NA
cnv	CNV parameters to keep in order to combine multi-omics data	NA	NA	NA
prot	Protein parameters to keep in order to combine multi-omics data	NA	NA	NA
variants_of_interest	takes a list of variants of interest in order to label data	["EIF6:20:33868702:T:C","TP53:17:7577559:G:T"]	NA
chr_of_interest	list of chromsomes to focus on it	["5","17","7"]	NA	any list of number of chromosomes

5.1. all_norm_dimred - snv

name	description	example	default value	possible value
method_dimred	reduction method to keep	pca	pca	fa,pca
dims	number of dimensions to keep	6	6	any integer
clustering_method	clustering method to keep	leiden	dbscan	graph-community,leiden,dbscan,hdbscan
res	resolution for clustering to keep	NA	NA	depend of the algorithm leiden and dbscan take float graph-community and hdbscan take integer
predict_missing_value	boolean to predict missing value using KNN method	True	False	True/False

5.2. all_norm_dimred - cnv

name	description	example	default value	possible value
method_dimred	reduction method to keep	pca	pca	pca
dims	number of dimensions to keep	6	6	any integer
clustering_method	clustering method to keep	leiden	dbscan	graph-community,leiden,dbscan,hdbscan
res	resolution for clustering to keep	NA	NA	depend of the algorithm leiden and dbscan take float graph-community and hdbscan take integer

5.3. all_norm_dimred - prot

name	description	example	default value	possible value
normalization	normalization method to keep	CLR	CLR	CLR,DSB,asinh,NSP
method_dimred	reduction method to keep	pca	pca	fa,pca
dims	number of dimensions to keep	6	6	any integer
clustering_method	clustering method to keep	leiden	dbscan	graph-community,leiden,dbscan,hdbscan
res	resolution for clustering to keep	NA	NA	depend of the algorithm leiden and dbscan take float graph-community and hdbscan take integer

6. phylogeny

name	description	example	default value	possible value
phylogeny_method	list of method to use for mutations events reconstruction	["COMPASS","infSCITE"]	NA	COMPASS,infSCITE,BiTSC2

6.1 phylogeny - COMPASS

name	description	example	default value	possible value
bool_cnv	add CNV in the reconstruction mutations events	1	0	0,1

What's coming next ?

---

• infSCITE and BiTSC2 are not implemented yet but they will be added soon
• Currently the version of mosaic used is 2.4.1, it will be updated to the 3.0.1

Questions

---

Don't hesitate to contact the bioinformatic plateform at bigr@gustaveroussy.fr or Remy.JELIN@gustaveroussy.fr if you have any questions/suggestion.