Authors: Cheng-Lin Frank Li ([email protected], [email protected])
Date: January 7, 2016
These scripts written in R are genetic variant annotation tools for Dictyostelium. It is designed for filtering and annotating the variant calling output (.vcf file) of the GATK UnifiedGenotyper (version 3.4-46). The README file contains guidelines for 1) how to prepare the Variant Call Format (VCF) file and 2) examples of applying the scripts to VCF files. Be aware of that the tools are designed for the VCF file format version 4.1 and the format of the VCF file are subjected to changes. If you encounter any error, feel free to send me an email.
This section will walk you through the steps and tools to generate a VCF file (.vcf) that is compatible with these scripts. If you create a VCF file by using different tools, you may encounter unexpected errors. For more information about the VCF file, please read the document.
1. Align the sequence reads
First, you will align the whole-genome sequence data (.bam files) to the reference genome (mask the duplication in the chromosome 2 from 3016083 to 3768654) by the BWA (0.7.5a). The chromosome 2 duplication region will be addressed later. If you use Bowtie to align your sequence reads, you may have to run java -jar GenomeAnalysisTK.jar -T PrintReads -R reference.fa -I input.bam -o output.bam -rf ReassignOneMappingQuality -RMQF 255 -RMQT 60 to covert the quality score before variant calling.
2. Mark duplicated reads
You will use the MarkDuplicates in the picard-tools (version 1.119) to mark duplicated reads. These duplicated reads will later be excluded from the variant calling.
3. Add read group
The GATK requires to have read group information in BAM files. So you have to run the AddOrReplaceReadGroups in the picard-tools (version 1.119) to add read groups. Here is an example of the read groups.
| RGID | RGLB | RGPL | RGPU | RGSM |
|---|---|---|---|---|
| strain01 | aggless | illumina | 2x100bp | AX4 |
| strain02 | aggless | illumina | 2x100bp | mutant01 |
| strain03 | aggless | illumina | 2x100bp | mutant02 |
As you might have noticed, the RGID and RGSM names must be unique for different samples. The strain01 is the parental strain, AX4. The strain02 and strain03 are chemically mutagenized strains that carry unknown mutations. The RGSM will be the column names of individual strains in the VCF file. Please name the RGSM of your mutant strains with a common prefix (e.g. “mutant” in this example).
4. Variant calling
Lastly, running the UnifiedGenotyper on multiple samples of the same experiment to generate a single VCF file (Note1). You will generate two VCF files for SNVs and indels, respectively, across the whole genome excluding the chromosome 2 duplication. Here are examples of the commands for running GATK in MAC OS X (v10.10.5). If you use other system or cloud server, make sure you have the same values for the three parameters (-glm, -stand_call_conf and -ploidy).
Note1: Do not run separate variant calling on individual samples and combine the individual vcf files later.
The VCF file contain the single nucleotide variants (SNVs)
Calling SNVs for a haploid genome. You will later use the snv.R to filter and annotate the VCF file.
java -jar gatk_directory/GenomeAnalysisTK.jar -T UnifiedGenotyper -glm SNP -R reference_fasta_file -I input1.bam [-I input2.bam ...] -o snv.vcf -stand_call_conf 30 -ploidy 1
The VCF file contain the indels
Calling indels for a haploid genome. You will later use the indel.R to filter and annotate the VCF file.
java -jar gatk_directory/GenomeAnalysisTK.jar -T UnifiedGenotyper -glm INDEL -R reference_fasta_file -I input1.bam [-I input2.bam ...] -o indel.vcf -stand_call_conf 30 -ploidy 1
1. The reference files
There are six reference files providing essential information for annotating variants (download the Reference_files.zip).
- gene_09-26-2015.txt
This file provides the gene model. It looks like this:
chromosome start end strand ddb_g genename
chr1 1890 3287 + DDB_G0267178 DDB_G0267178_RTE
- dicty_primary_cds.txt
This file provides the dictybase ID and the coding sequence of each gene. It looks like this:
DDB0216524|DDB_G0267364
"a" "t" "g" "g" "t" "c" "g" "t" "a" "c" "c" "a" "g" "a" "c" "t" "g" "g" "t" "t" "a" "c" "a" "a" "a" "c" "t" "g" "a" "t" "g"
....
- cds_09-26-2015.gff
This file is a General Feature Format (GFF) file of coding DNA sequence (CDS). It looks like this:
seqname source feature start end score strand frame attribute chromosome
DDB0232428 Sequencing Center CDS 1890 3287 . + . Parent=DDB0216437 chr1
- splice_acceptor.gff
This file is a General Feature Format (GFF) file of splicing acceptor sites of the introns. It looks like this:
seqname source feature start end score strand frame attribute chromosome n
DDB0232428 Sequencing Center CDS 11262 11263 . + . DDB0216442 chr1 3
- splice_donor.gff
This file is a General Feature Format (GFF) file of splicing donor sites of the introns. It looks like this:
seqname source feature start end score strand frame attribute chromosome n
DDB0232428 Sequencing Center CDS 11200 11201 . + . DDB0216442 chr1 3
- dd_codon_bias.txt
This file provides the frequency of codon usage.
Codon AA Number freq freq_abs
GCA Ala 109859 1.67 1.668609801
2. Applying the scripts to your VCF files
You will find the R scripts in the folder /R, exmaple files in the folder /examples and the raw vairant calling results from genetic screens performed in Dictyostelium in the folder /data.
Here is an exmaple of applying the snv.R in the folder /R to filter and annotate the snv.vcf in the folder /examples. The VCF file contains 39 SNVs as a result of whole-genome SNVs calling. First, let's take a look of the snv.vcf. If you are not familiar with the format, please read the document
#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT AX4 mutant01 mutant02 mutant03 mutant04 mutant05 mutant06
chr6 773798 . A G 5181.7 . AC=1... GT:AD:DP:GQ:PL 0:68,0:68:99:0,2891 ...
The parental strain is the "AX4". Mutant strains is named "mutant01" to "mutant06". To filter and annotate the SNVs. You will run the following commands in R.
source("/path_to_the_folder/snv.R")
snv(input_file = "/path_to_the_folder/snv.vcf", ref_file = "/path_to_the_folder_of_reference_files/", parental_strain = "AX4", mutant_strain = "mutant")
The script creates a new folder /snv.vcf_5. The suffix 5 indicates that SNVs with a coverage < 5 reads will be filtered out. You can change the threshold by adding the argument read_depth = 5 in the function snv(). The script generates an output folder containing 8 files.
| File name | Description |
|---|---|
| summary.txt | Summarize the input parameters and results. |
| variants.vcf | A genome VCF file of variants that passed the filters. You can view this file in the Integrative Genomics Viewer (IGV). |
| variant_filtered.txt | A data frame of variants that passed the filters. |
| variant_mult_alt.txt | A data frame of variants that contain more than two alternative alleles. These vairants are likely to be false positives. |
| gene_list_all.txt | Genes that are mutated at least once. |
| gene_list_top.txt | Genes that are mutated at least twice. |
| mutations_by_chr.txt | List mutations in individual chromosomes. |
| mutations_by_strain.txt | List mutations in individual strains. |
| strain_by_gene.txt | List mutants that carrying mutations in individual genes. |
The summary.txt looks like this:
input_file = ...
parental_strain = AX4
mutant_strain = mutant
...
88: total number of variant sites before filtering
87: remaining variants after removing sites where the parental strain has no read (filter 1).
86: remaining variants after filter 2 (genotype filter). Allowed genotypes : ., 0, 1.
66: remaining variants after filter 3.1 (frequency filter)
42: remaining variants after filter 3.2 (freq_diff & read_depth filters).
1: variants with multiple alternative alleles that pass filter 3.2 (freq_diff & read_depth filters).
...
In brief, there are 88 SNVs before filtering. One SNVs did not pass the filters because it contains multiple alternative alleles. 42 SNVs pass all filters (36 SNVs are found in one mutant and 3 SNVs are found in two mutants, 36 + 3*2 = 42). The 42 SNVs are listed in the variant_filtered.txt. The results can be found in the folder /snv.vcf_5.
Similarly, you can apply the indel.R in the folder /R to filter and annotate the indel.vcf in the folder /examples. The indel.vcf contains 90 indels but only 2 of them pass the filters. No variant with multiple alternative alleles are found. The results can be found in the folder indel.vcf_5.
In the variant_filtered.txt, description of the columns can be found in the following table. The symbol NA in the file means a value is not available.
| Column name | Description |
|---|---|
| CHROM | Chromosome location. |
| POS | Nucleotide position of the variant. |
| REF | Reference allele. |
| ALT | Alternative allele. |
| QUAL | Quality score. Not a useful value. |
| MQ | Mapping Quality: root mean square of the mapping quality of reads across all samples. |
| QD | Quality by Depth: variant confidence normalized by unfiltered depth of variant samples. |
| Strain id | The name of the chemically-mutagenized strain that carries the variant. |
| DDB_G | Dictybase gene ID. |
| Genename | Gene name. |
| Description | Description of the gene. |
| Start | The start position of the gene. |
| End | The end position of the gene. |
| Strand | The gene is located on forward(+) or reverse(-) strand. |
| dna_change | Alteration in the DNA sequence. |
| codon_dna | DNA codon of the reference allele. |
| codon_dna_mut | DNA codon of the mutant allele. |
| codon_freq | Codon usage frequency of the reference DNA codon. |
| codon_freq_mut | Codon usage frequency of the mutant DNA codon. |
| codon_pro | Amino acid residue of the wildtype protein. |
| codon_pro_mut | Amino acid residue of the mutant protein. |
| aa_pos | Amino acid position where mutation occured. |
| splice_variant | Sequence ID of the splice variant. |
| other_splice_variants | Sequence ID of alternative splice variants. |
| mutation | Mutation type. Intergenic, intronic, splice_donor, splice_acceptor, synonymous, missense, nonsense or non-coding gene (NCG). |
In the Part I & II, we treat the Dictyostelium genome exluding the chromosome 2 duplication as a haploidy. In Part III, we will do variant calling, filtering and annotation in the chromosome 2 duplication region by considering it as a diploidy. All the steps are basically the same as the Part I & II. The differences are adding the arguments -ploidy 2 -L chr2:2263132-3015703 during vairant calling and use the R scripts (snv_chr2.R and indel_chr2.R) for filtering and annotation.
1. Variant calling
The VCF file contains single nucleotide variants (SNVs) in the chromosome 2 duplication
Calling SNVs in the chromosome 2 duplication region. You will later use the snv_chr2.R to filter and annotate the VCF file.
java -jar gatk_directory/GenomeAnalysisTK.jar -T UnifiedGenotyper -glm SNP -R reference_fasta_file -I input1.bam [-I input2.bam ...] -o snv_chr2.vcf -stand_call_conf 30 -ploidy 2 -L chr2:2263132-3015703
The VCF file contains indels in the chromosome 2 duplication
Calling indels in the chromosome 2 duplication region. You will later use the indel_chr2.R to filter and annotate the VCF file.
java -jar gatk_directory/GenomeAnalysisTK.jar -T UnifiedGenotyper -glm INDEL -R reference_fasta_file -I input1.bam [-I input2.bam ...] -o indel_chr2.vcf -stand_call_conf 30 -ploidy 2 -L chr2:2263132-3015703
No example is provided for this part.