MosaiC-All

Comprehensive analysis of somatic and parental gonosomal mosaicism using trio and singleton design.

Step 1: Pre-requisites/ Softwares Installation

The following are softwares and resources that can be dowloaded/prepared based specified instructions.

1.1 Downloadable sources

MosaicHunter: https://github.com/zzhang526/MosaicHunter
MosaicForecast: https://github.com/parklab/MosaicForecast
GATK: https://github.com/broadinstitute/gatk/releases

1.2 General Resources

Resources	Example/Sources/Notes
Reference genome	e.g hs37d5.fa
Common variant databases	dbSNP (e.g b37_dbsnp_138.b37.vcf) gnomAD (e.g somatic-b37_af-only-gnomad.raw.sites.vcf)
Repeats	e.g all_repeats.b37.bed; can be found in MosaicHunter installation (i.e $MHDIR/MosaicHunter-master/resources)
Exome errors databases	e.g WES_Agilent_71M.error_prone.b37.bed; can be found in MosaicHunter installation (i.e $MHDIR/MosaicHunter-master/resources)
Panel of Normal (PON)	Should be prepared based on samples that are not part of the analysis. As a suggestion for large cohort analysis, samples can be divided into two cohorts to create two Panel of Normals (PON_A and PON_B). This PON can be prepared based on GATK option-CreateSomaticPanelOfNormals (i.e https://gatk.broadinstitute.org/hc/en-us/articles/4405451431963-CreateSomaticPanelOfNormals-BETA)

Step 2: Config-files

There is a config-file, in which directories of softwares/resources (as prepared in Step 1) should be specified.

Mosaic-All.config

Step 3: Variant calling using four tools

3.1 Summary:

Variants are either detected in singletons and/or trios using three mosaic variant callers (MosaicHunter, MosaicForecast, Mutect2) and GATK-HC. A wrapper-script (MasterScript_MosaiC-All.sh) was developed for this purpose, and can be executable in University of Adelaide HPC environment using the following command.

3.2 Command:

MosaiC-All/MasterScript_MosaiC-All.sh -s $SampleID.list -o $Outputs -c MosaiC-All/Mosaic-All.config

3.3 Requirements:-

1. $SampleID.list: A tab-separated-file as following format based on the Bam.files of each sample (e.g 001P.realigned.bam)

Directory of Bam files	ProbandID	Gender	MotherID	FatherID
./path	001P	F	001M	001F

2. $Outputs: An Output directory to store all final outputs

3. $DIR/MosaiC-All/MosaiC-All.config: A config file prepared as mentioned in Step 1 and 2.

Step 4: Analysis for somatic mosaicism (M3 pipeline)

4.1 Processing all outputs

Aims:

• to filter MFcalls manually and
• Merge all variants based on each tool.

sbatch $SCRIPTDIR/CombineCalls.sh -s $sampleID -d $Outputs

Requirements:-

• sampleID (i.e 001P)
• Outputs (Output directory as specified in Step 3)

4.2 Find overlaps using R script (M3pipeline.R)

Aims:

• To flag variants that were found in same sample, using 1-3 mosaic variant calling tools
• Followed by filtering out variants that were found only by one tool

Step 5: Analysis for parental gonosomal mosaicism (pGoM)

Aims:

• To filter parental mosaic variant calls based on transmission to children

5.1 Prefilter

• Identify inherited variants that expected to be mosaic based on AAF and GT, using GATKHC outputs

Command:

sbatch /MosaiC-ALL/postprocessing/pGoM.sh -v /path/to/directory_of_vcf -s FamilyID.txt -o /path/to/output/directory

Requirements:

1. input directory (where can we find the family.vcf).

2. sampleID list (one header row and then tab-delimited columns $BAMdir,$ProbandID,$Gender,$Mother,$Father).

Directory of Bam files	ProbandID	Gender	MotherID	FatherID	FamilyVCF
./path	001P	F	001M	001F	Trio001.vcf
./path	004P	F	004M	004F	004.family.vcf

3. /path/to/output/directory (A location for the output files).

5.2 Postfilter

• Mosaic variants are identified among prefiltered pGoM variants using one or more mosaic variant calling tools
• Example script: MosaiC-ALL/postprocessing/pGoMpipeline.R

Requirements:

1. Three Output files from MosaiC-ALL/postprocessing/M3_CombineCalls.sh
2. pGoM.sh output file
3. Amend the working directory and output_file prefix in the R.script