How to tune GATK for low-coverage data?

Introduction to GATK and Low-Coverage Data

• GATK (Genome Analysis Toolkit) is a toolkit developed to manage next-generation sequencing data.

 

• Low-coverage data refers to sequencing data that has insufficient read coverage across the genome.

Initial Setup and Environment

• Ensure that your environment has the appropriate version of GATK installed with all necessary dependencies.

 

• Prepare your dataset. For low-coverage data, ensure quality control steps such as adapter removal and base quality trimming have been thoroughly conducted.

Pre-Processing Steps

• Align the low-coverage reads to a reference genome using an aligner like BWA-MEM.

 

• Sort and index the aligned BAM files using tools like Samtools.

 

• Mark duplicates in your BAM files using GATK's `MarkDuplicates` to reduce variability caused by PCR duplication.

Handling Low-Coverage in GATK

• Use `HaplotypeCaller` in the `ERC` (GVCF) mode to call variants. This method allows joint genotyping for more accurate variant calling.

 

• Adjust the `--min-base-quality-score` if necessary, based on the quality of your reads, to ensure low-quality bases do not skew results.

 

• Tune `--standard-min-confidence-threshold-for-calling` to manage the confidence level for calling variants in low-coverage regions.

 

• Review the `--min-pruning` parameter which helps omit branches in the assembly process, especially useful in sparse data.

Post-Procesing and Analysis

• Realign around indels to correct misalignments caused by insertion or deletion polymorphisms using `RealignerTargetCreator` and `IndelRealigner`.

 

• Perform base quality score recalibration with `BaseRecalibrator` and `ApplyBQSR` for a more accurate base quality assessment.

 

• Conduct variant quality score recalibration (VQSR) if sufficient variant data is available or use hard filtering for very low-coverage data sets.

Validation and Quality Control

• Compare variant calls against a validated dataset if available, or use tools such as `VerifyBamID` to estimate contamination levels.

 

• Use `VariantEval` to assess the quality of the variant calls by analyzing metrics such as Ti/Tv Ratio, Depth of Coverage, etc.

 

• Consider integrating externally available tools or benchmarks like Genome in a Bottle (GIAB) for accuracy assessment.