# DRAGEN Array v1.4.0 Release Notes

## **RELEASE DATE**

May 2026

## **RELEASE HIGHLIGHTS**

* **New QC Report capability** enabling comprehensive Infinium Array sample QC review via an interactive HTML report, including:
  * Control Dashboard (replacing the GenomeStudio control dashboard)
  * Automated QC
  * Sample QC Heatmaps
  * Trend Analysis
* **Major cytogenetics algorithm and feature enhancements**, including new GAINLOH calling support and improved mosaic detection.
* **PGx workflow improvements**, encompassing structural variant star‑allele calling, enhanced handling of variant and star‑allele no‑calls, and redesigned output formats.
* **Improved multi‑allelic variant (MAV) calling**, with more robust handling of Infinium I no‑call scenarios.

## **NEW FEATURES IN DETAIL** <a href="#section-v14-new-features-in-details" id="section-v14-new-features-in-details"></a>

* Cytogenetics
  * Added support for detecting GAINLOH (Gain with Loss of Heterozygosity) variants, which are genomic regions characterized by increased copy number (gain) and loss of heterozygosity (LOH).
  * Added variant size and probe filters for each variant type (e.g., deletion, duplication, LOH, GAINLOH, mosaic) to improve flexibility.
  * Enhanced mosaic variant detection. The algorithm detects mosaicism at approximately >15% for mosaic states with a one-copy difference (e.g., CN2/CN1, CN2/CN3 in diploid regions; CN1/CN0, CN1/CN2 in haploid regions).
  * Support detection of mosaic duplication on the male Y chromosome.
  * CNV and LOH variants are validated against expected LRR patterns (e.g., most probes in an autosomal CN gain region should exhibit positive LRR); variants that do not conform are filtered out.
  * Fixed an issue in the cyto annotation JSON file where the first variant was missing ISCN annotation.
  * Removed the 5% mosaic fraction hard cutoff so variants below this threshold can still be reported.
  * LOH regions smaller than 500 kb that are flanked by copy-neutral regions are removed to reduce false positives.
* QC Report
  * Generates a comprehensive QC report to support quality control review of genotyping and DNA methylation microarray datasets processed with DRAGEN Array.
  * Enables review of sample‑level and dataset‑level QC metrics, including control metrics and plots, to identify data quality issues and outliers.
  * Supports single‑dataset QC in the cloud; the local CLI additionally supports multi‑dataset analysis and cross‑run summary views.
  * Provides interactive visual dashboards, including Control Dashboard, Automated QC, Sample QC Heatmaps, and Trend Analysis.
  * Output includes sample QC table and interactive HTML report.
  * Supports configurable QC thresholds via YAML configuration files and in‑report QC Metric Configuration.
  * For usage instructions, input requirements, and output details, see the [QC Report documentation](/product-guides/dragen-array-local-analysis/qc-report.md).
* PGx
  * Improved PGx CNV calling robustness through algorithm updates that reduce sensitivity to outlier samples.
  * Enhanced star‑allele calling accuracy via targeted algorithm improvements across supported PGx genes.
  * Clear separation of no‑call conditions, explicitly distinguishing "Per‑sample variant no‑calls" and "Variants not represented on the array product". These are now reported separately, with affected star alleles explicitly identified.
  * Ranked candidate genotypes for no‑call results, reporting the closest star‑allele solutions along with variant delta annotations that describe differences between the sample input and each candidate genotype.
  * Expanded variant traceability by including RSIDs and probe IDs in both supporting and no‑call variant outputs, facilitating data review, troubleshooting, and audit workflows.
  * Direct reporting of PGx annotations in CSV outputs, including: genotype, associated metabolizer status, PGx guideline applied.
  * Redesigned CSV reporting architecture to simplify downstream parsing, improve auditability, and support custom reporting workflows. Output is now organized into multiple clearly defined files, each serving a distinct purpose.
  * Fixed rare intermittent memory issues occurring during star‑allele calling.
  * Corrected CYP2D6 exon 9 conversion reporting, resolving an issue where non‑\*36 star alleles (e.g., \*83) were incorrectly reported as \*36 with an underlying \*83 allele.
* Genotyping & Core
  * Sample name used in downstream outputs now follows a defined precedence: See [Sample Name Determination](/product-guides/input-files.md#sample-name-determination).
  * The genotyping cloud pipeline now outputs QC metrics and the QC report.
  * Improved accuracy for MAV calling via the `--use-infI-nc-info` option in [genotype gtc-to-vcf](/product-guides/dragen-array-local-analysis.md#section-genotype-gtc-to-vcf)

## **BUG FIXES**

* Genotyping & Core
  * The [samplesheet](/product-guides/input-files.md#section-sample-sheet) now handles empty columns.
  * During locus combination logic for `gtc-to-vcf`, duplicate INDELS with HET calls on opposite strands (i.e., probe calls of `D/I` + `I/D`) now correctly report HET (`0/1`) in the VCF instead of a no call (`./.`).

## **IMPORTANT NOTES**

* DRAGEN Array v1.4 is a self-contained .NET 9 application. While the .NET runtime is bundled with the application, Linux users may need to install native OS-level dependencies. See [Linux Dependencies](/product-guides/dragen-array-local-analysis.md#linux_dependencies) in the installation guide for details.

## **KNOWN ISSUES** <a href="#section-v14-known-issues" id="section-v14-known-issues"></a>

* PGx
  * There can be some minor differences when running `pgx star-allele call` on Windows vs. Linux. See [DRAGEN Array v1.3.0 Release Notes](https://github.com/illumina-swi/dragen-array-docs/blob/DAv1.4/docs/reference/release-notes/dragen-array-v1.3.0-release-notes). All overall solutions tested for comparison were found to be concordant.
  * Occasional star-allele solution unresolved variants discordance between Linux and Windows OS with concordant solution ranking.
  * For PGx star-allele solution no calls, the PGx caller may return alternate closest solutions with a large variant edit distance >= 3 when low quality variant calls are present in the input sample.
  * Some simple variants have REF and ALT delimited by `_` instead of `>` in the `star_alleles.csv` and metabolizer status JSON files (e.g., `ryr1.38577931a_c` instead of `ryr1.38577931a>c`)
  * The JSON output for star-allele candidate solutions includes supporting details for only the first allele in the candidate diplotype. This issue is limited to the JSON file; complete supporting information is correctly reported in the CSV outputs.
* Genotyping & Core
  * Corrupt or invalid GTC files will abort with an error instead of skipping. The corrupt or invalid GTC files will need to be removed before proceeding.
  * Some multi-nucleotide variant (MNV) designs reverse complement the "Allele1/2 Top" fields in the Final Report
  * GTC files do not support non-ASCII characters. This is especially problematic when running DRAGEN Array local if operating system locale settings are not English-based (e.g., en-US) as internal datetime fields could write non-ASCII characters. This will result in the following error:

    ```
    fail:  ArrayAnalysis.CLI.App[0] 
            [07:17:07 6620]: System.IO.EndOfStreamException: Unable to read beyond the end of the stream. 
                 at System.IO.BinaryReader.ReadString() 
                 at ArrayAnalysis.Core.GtcFileLoader..ctor(String filePath) in /src/ArrayAnalysis.Core/GtcFileLoader.cs:line 161 
                 at ArrayAnalysis.Services.GtcFactory.CreateGtcFromSample(Sample sample, Boolean log) 
                 at ArrayAnalysis.Services.GtcToVcfService.Run(GtcToVcfInput input) 
                 at ArrayAnalysis.CLI.App.RunCliServiceAndReturnExitCode(BaseOptions opts) in /src/ArrayAnalysis.CLI/App.cs:line 110
    ```

    There is a workaround to disable globalization and produce valid GTC files:

    1. Locate the `dragena.runtimeconfig.json` file inside the installation directory of DRAGEN Array (i.e., where the .zip or .tar.gz file was downloaded and unzipped).
    2. Add the key `System.Globalization.Invariant` to that file and set its value to `true`. (i.e., step #2 here: <https://github.com/dotnet/corefx/blob/master/Documentation/architecture/globalization-invariant-mode.md#enabling-the-invariant-mode>)
    3. Re-generate the GTC using the `genotype call` subcommand.
  * SNV and indel variants are always treated as separate variants and are not collapsed in gtc-to-vcf even if they are designed to the same locus.
  * Some indel variants are missing from SNV VCF due to mapping issue between the designed indels and the reference genome.
  * In the gtc-to-vcf subcommand a mismatch between BPM and CSV manifests will not cause the command to abort with an error. The mismatch will need to be addressed before proceeding.
  * Running `genotype call` or `qc call` with mismatched IDATs and manifest files could produce inaccurate results, but the software will not produce any errors. Users should double-check they are using the correct manifest file for their array if genotype call rates are unexpectedly low and/or QC values have many NaN values.
  * Very large samplesheets (i.e., >1K samples) can significantly slow analysis. The recommended workaround is to split analyses into batches of at most 1K samples.

## **KNOWN LIMITATIONS** <a href="#section-v14-known-limitations" id="section-v14-known-limitations"></a>

* Cytogenetics
  * If the genotyping module reports an unknown sex and the cytogenetic caller cannot resolve it, the caller assumes the sample is male. As a result, sex chromosome detection may be inaccurate if the sample is actually female. This behavior is not currently output in the log.
  * ISCN annotations in the cytogenetic annotation JSON output file are only provided for variants greater than 1 kb in length. This is often cited as a minimum size limit used to define copy number variants.
  * Centromere regions typically have low sequence complexity and are prone to artifacts. As a result, cytogenetic calling results in these regions may possibly be false positives.
  * ISCN annotations are not fully supported for LOH and mosaic variants in the cytogenetic annotation JSON output file.
  * DRAGEN Array Cytogenetics analysis is intended for constitutional samples only, oncology samples are not supported at this time.
  * DRAGEN Array Cytogenetics analysis is validated only for specific array platforms: Infinium Global Diversity Array with Cytogenetics-8, Infinium Global Screening Array with Cytogenetics-24, and Infinium CytoSNP-850K BeadChip (iScan and NextSeq550).
  * The Cytogenetics algorithm assumes that most genomic regions are diploid for optimal performance. As a result, variants in triploid samples may be missed.
  * DRAGEN Array Cytogenetics analysis may call large events that are broken into smaller pieces and require visual confirmation.
  * GT is hardcoded to homozygous alt (1/1) for cyto VCF entries.
  * DRAGEN Array Cytogenetics analysis does not produce mosaic ISCN notation at this time.
  * Older Cyto model files and cloud configurations are not compatible with v1.4.
* QC Report
  * Hover text in the Control Dashboard overall box plot is truncated at 40 characters, which may hide additional columns.
  * For very large datasets (many plates or barcodes), buttons in the heatmap menu may appear misaligned.
  * Faceting behavior groups categorical levels beyond six into an "Other" category rather than displaying all levels individually.
  * In the Control Dashboard, after viewing a dataset with more than 12,000 samples, you may be unable to switch to a dataset with fewer than 1,000 samples; if this occurs, reload the HTML report.
  * In the Control Dashboard, once samples are selected, moving or resizing the selection can cause the HTML interface to become unresponsive and the operation may take a long time to finish.
* PGx
  * Command line options `unsquash-duplicates` and `filter-loci` for `gtc-to-vcf` conversion should not be used when star allele calling is desired. In addition, VCFs must be gzipped and tabix indexed (the default for `gtc-to-vcf`) to be used in star allele calling.
  * Star allele calling does not support novel alleles; only alleles defined in the PharmVar and PharmGKB databases are supported.
* Genotyping & Core
  * Genotyping only supports diploid organisms. Polyploid genotyping is currently not supported.
  * Tabix indexing from DRAGEN Array is not exactly the same as [bcftools index --tbi](https://samtools.github.io/bcftools/bcftools.html#index). For instance, if you run `bcftools index --stats in.vcf.gz` or `bcftools index --nrecords in.vcf.gz`, with certain versions of bcftools, you may get the following error: `index of in.snv.vcf.gz does not contain any count metadata. Please re-index with a newer version of bcftools or tabix.`. If these tools are critical to user's bioinformatics pipelines a workaround would be to unzip and re-index DRAGEN Array VCFs using bcftool's tabix. But please note, these index files may not work in downstream VCF-based DRAGEN Array commands like `pgx star-allele call`. Please use DRAGEN Array end-to-end for analysis flows like the ones detailed in the [Quick Start](/product-guides/dragen-array-local-analysis.md#section-quick-start) guide.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://help.dragenarray.illumina.com/support-and-updates/release-notes/dragen-array-v1.4.0-release-notes.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.