Variant Filters Using Segregation Information

This repository contains the code used for the study: "Variant Filters Using Segregation Information Improve Mapping of Nectar-Production Genes in Sunflower (Helianthus annuus L.)". The study evaluates the impact of biologically informed variant filtering strategies on QTL mapping, demonstrating improved identification of candidate genes related to nectar production.

---

Contents

CandidateGeneGetter.sh

This shell script extracts candidate genes from a GFF annotation file (HAN412_Eugene_curated_v1_1.gff3) based on genomic regions specified in the Windows file. For each region (defined by chromosome, start position, and end position), it identifies all genes falling entirely within that window, counts them, and outputs the region information along with a comma-separated list of gene IDs to AshleyCandidateGenes.txt.

Chi_square_template.R

This R script filters genomic markers using a chi-square test based on expected segregation ratios. The script is designed as a template that can be adjusted for different population types by modifying the expected ratios. The default values (48.4375% homozygous for each allele and 3.125% heterozygous) are set for F6 inbred lines, but can be modified to match the segregation expectations of any population being filtered. It retains markers whose observed genotype frequencies do not significantly deviate from expectations (p > 0.1), removing markers with segregation distortion that could interfere with accurate QTL identification.

GATK_TopDown_Parallel_Pipeline.sh

This shell script implements a comprehensive, parallelized pipeline for variant calling, starting from raw FASTQ files and producing a final multi-sample VCF. It is designed for processing numerous samples efficiently by leveraging GNU Parallel across genomic regions defined in a BED file (pathToBED). The script requires specific paths to executables (fastp, bwa-mem2, samtools, gatk, parallel) and reference files, as well as adherence to FASTQ naming conventions. It relies on modules loaded in an HPC environment (gatk, parallel, samtools, bcftools). Key steps include:

1. Read Processing & Alignment: Performs quality control and adapter trimming using fastp, followed by alignment to a reference genome (pathToReference) using bwa-mem2. The resulting alignments are processed (collated, fixed, sorted, duplicates marked) using samtools to generate analysis-ready BAM files for each sample (${sample}.sorted.bam). The script iterates through samples listed in a file named ToRedo.
2. Per-Sample Variant Calling: Runs gatk HaplotypeCaller in GVCF mode in parallel across genomic regions (defined in pathToBED) for each sample. It then concatenates the resulting regional GVCFs into a single per-sample GVCF (concat_fromBAM_${sample}.g.vcf.gz) using bcftools.
3. GVCF Aggregation (Commented Out): Prepares for joint genotyping by combining per-sample GVCFs into a GenomicsDB using gatk GenomicsDBImport. (Note: This step is commented out in the provided script).
4. Joint Genotyping: Performs joint variant calling across all samples using gatk GenotypeGVCFs on the GenomicsDB (created in the previous, potentially commented-out step), run in parallel across genomic regions. The resulting regional VCFs are concatenated using bcftools into a final multi-sample VCF (Combined.vcf.gz). * The final output of the active pipeline steps is the merged, unfiltered multi-sample VCF file named Combined.vcf.gz.

mapping.R

This R script performs QTL (Quantitative Trait Locus) mapping using the qtl package. It includes code for three distinct "Approaches," likely representing analyses performed on different datasets or using varied marker filtering strategies (Approach1.csv, Approach2.csv, Approach3.csv). The script covers data loading, genetic map estimation and refinement (including custom marker thinning functions and visualization of recombination frequencies), calculation of genotype probabilities, performing 1D (scanone), Composite Interval (cim), and 2D (scantwo) QTL scans, significance testing via permutations, and refining QTL models (fitqtl, refineqtl).

marker_filt_dist.R

This R script filters genomic markers from a VCF file by removing markers within 125,000 bp of each other. It optimizes marker density while maintaining genome-wide coverage, ensuring the filtered set is suitable for QTL mapping and identifying genomic regions linked to nectar-production traits in sunflower.

proc freq marker data.sas

This SAS script filters genetic markers based on segregation patterns. It utilizes PROC FREQ to calculate genotype frequencies for biallelic markers (assuming three genotype classes) and performs chi-square tests against expected segregation ratios (e.g., specified test probabilities like 0.484375, 0.03125, 0.484375, corresponding to F6 expectations). Markers significantly deviating from these expectations (p < 0.10 in this script) are identified and potentially excluded from downstream analyses, similar in principle to Chi_square_template.R but implemented within the SAS environment for specific datasets (markers.bialw).

thinning_loop.R

This R script thins genomic markers based on inter-marker distance thresholds, identifying and removing redundant or closely spaced markers. It helps refine marker sets to balance genome coverage and computational efficiency, improving QTL mapping precision in the study of sunflower nectar-production traits. (Note: Similar custom functions are also included within mapping.R).

Windows

This plain text file serves as input for the CandidateGeneGetter.sh script. Each line defines a genomic window with three columns: Chromosome, Start Position, and End Position. These windows likely represent regions of interest identified through QTL mapping or other analyses.

---

Citation

Barstow, A.C., McNellie, J.P., Smart, B.C., Keepers, K.G., Prasifka, J.R., Kane, N.C., & Hulke, B.S. (2025). Variant filters using segregation information improve mapping of nectar-production genes in sunflower (Helianthus annuus L.). The Plant Genome.