README.md

## Introduction

**Whole Exome** data analysis - from fastq data to variants calling and annotation
As stated in supplementary methods we performed WES analysis relying mostly on the GATK best practices for WES analysis.  
Raw fastq data are available at [PRJEB63520](https://www.ebi.ac.uk/ena/browser/view/PRJEB63520) repository.   

Below the main steps performed and the relative running commands:

# Alignment #

Fastq files were aligned with [BWA aligner](https://github.com/lh3/bwa) (v0.7.17) to GRCh38 reference genome (GRCh38.p13 gencodegenes) using default parameters, except for the -M option for [Picard](https://broadinstitute.github.io/picard/) compatibility necessary for marking of duplicates.  

```
- bwa mem -t 12 -R @RG\tID:SampleID_L1\tSM:SampleID  PL:ILLUMINA -M GRCh38.p13.genome.fa 3_S8_L001_R1_001.fastq.gz 3_S8_L001_R2_001.fastq.gz` 
- picard SortSam INPUT=SampleID_L1.sam OUTPUT=BALL_12_27_1_L1_mouse.bam SORT_ORDER=coordinate # both lanes L1 and L2
- picard MergeSamFiles I=SampleID_L1_mouse.bam I=SampleID_L2_mouse.bam OUTPUT=SampleID_mouse.bam
- samtools index SampleID_mouse.bam
- picard MarkDuplicates INPUT=SampleID_mouse.bam OUTPUT=SampleID.dedup_reads_mouse.bam METRICS_FILE=SampleID.metrics_mouse.txt
- gatk BaseRecalibrator --input SampleID.dedup_reads_mouse.bam --reference $genome --known-sites $vreference --output SampleID_recal_data_mouse.table
- gatk ApplyBQSR --reference $genome --input SampleID.dedup_reads_mouse.bam --output SampleID.dedup_reads_mouse_recal.bam --bqsr-recal-file SampleID_recal_data_mouse.table --static-quantized-quals 10 --static-quantized-qual
- gatk BaseRecalibrator --input SampleID.dedup_reads_mouse_recal.bam --reference $genome --known-sites $vreference --output SampleID_post_recal_data_mouse.table
```

## Disambiguation ##

We perform alignment using both human and mouse reference genomes in order to perform reads disambiguation using [disambiguate](https://pubmed.ncbi.nlm.nih.gov/27990269/) and discard reads from human mapping that belong to mouse cells.  

```
python disambiguate.py -a bwa -s SampleID 02-Alignment_human/SampleID.dedup_reads_rehead_recal.bam 02-Alignment_mouse/SampleID.dedup_reads_mouse_recal.bam
```
# Variant calling #

For variant calling analysis we opted for Mutect2 algorithm (more sensitive for somatic variants identification) in tumor-only mode.  
Below the workflow and commands used:  

```
- gatk Mutect2 -I SampleID.disambiguatedSpeciesA.bam.sorted.bam -O Mutect2_raw_call_SampleID_exome_refseq_light.vcf.gz -R GRCh38.p13.genome.fa -L exome_light.bed -G StandardMutectAnnotation -OVI true -OVM true -ip 50
- gatk FilterMutectCalls -R GRCh38.p13.genome.fa -V Mutect2_raw_call_SampleID_exome_refseq_light.vcf.gz -O Mutect2_raw_call_SampleID_exome_refseq_light_filtered.vcf.gz
```

# Annotation #

We implemented several layers of annotation using different datasets including [SNPeff](http://pcingola.github.io/SnpEff/) (v.86), [dbSNP](https://www.ncbi.nlm.nih.gov/snp/) (v.152), [dbNSFP4](http://database.liulab.science/dbNSFP) (4.0a for academic). 

```
snpEff -v -csvStats ${sampleid}\_rawstats_canon.csv -s ${sampleid}\_rawstats_canon.html -canon GRCh38.86 Mutect2_raw_call_${sampleid}\_exome_refseq_light_filtered.vcf.gz > $sampleid\_raw_annot_refseq_light_filtered.vcf

# dbSNP annotation

SnpSift annotate $dbsnp ${sampleid}\_raw_annot_refseq_light_filtered.vcf > ${sampleid}\_dbsnp_refseq_light_filtered.vcf
gatk IndexFeatureFile -I ${sampleid}\_dbsnp_refseq_light_filtered.vcf

# Cosmic annotation (coding)

SnpSift annotate $cosmic ${sampleid}\_dbsnp_refseq_light_filtered.vcf > ${sampleid}\_dbsnp_cc_refseq_light_filtered.vcf
gatk IndexFeatureFile -I ${sampleid}\_dbsnp_cc_refseq_light_filtered.vcf

# Cosmic annotation (non coding)

SnpSift annotate $cosmicnc ${sampleid}\_dbsnp_cc_refseq_light_filtered.vcf > ${sampleid}\_dbsnp_cc_cnc_refseq_light_filtered.vcf
gatk IndexFeatureFile -I ${sampleid}\_dbsnp_cc_cnc_refseq_light_filtered.vcf

# dbNSFP4 annotation 

SnpSift dbnsfp -db dbNSFP4.0a/dbNSFP4.0a.txt.gz -v ${sampleid}\_dbsnp_cc_cnc_refseq_light_filtered.vcf  > ${sampleid}\_dbsnp_cc_cnc_dbnsfp_refseq_light_filtered.vcf

```

# Field extraction for AF comparison #

We leverage SnpSift extractfield function for selecting fields of interest and performing Allele Fraction analysis.

```
- SnpSift extractFields ${sampleid}_dbsnp_cc_cnc_dbnsfp_refseq_light_filtered.vcf CHROM POS REF ALT GEN[0].AF GEN[0].DP GEN[0].AD[0] GEN[0].AD[1] GEN[0].GT ID FILTER ANN[0].HGVS_P ANN[0].GENE ANN[0].BIOTYPE ANN[0].RANK ANN[0].EFFECT ANN[0].IMPACT COMMON G5 dbNSFP_ExAC_Adj_AF dbNSFP_1000Gp3_AF dbNSFP_ExAC_AF dbNSFP_phastCons100way_vertebrate dbNSFP_FATHMM_pred dbNSFP_GERP___RS dbNSFP_GERP___NR dbNSFP_CADD_phred dbNSFP_MetaSVM_pred dbNSFP_LRT_pred dbNSFP_PROVEAN_pred dbNSFP_MutationTaster_pred dbNSFP_MutationAssessor_pred dbNSFP_SIFT_pred dbNSFP_Polyphen2_HVAR_pred dbNSFP_Polyphen2_HDIV_pred > ${sampleid}_dbsnp_cc_cnc_dbnsfp_refseq_light_raw_filtered.fields.txt
```