v1.0.2

[1.0.2] (11 June 2025)

This release includes improvements to Poly(A/T) estimation.

• cd6d5b2 - Improved split Poly(A/T) tail detection for plasmids.
• 2df0f52 - Allow use of old-style calibration_coefficients to support poly(A/T) calling for legacy models.

v1.0.1

[1.0.1] (4 June 2025)

This release introduces support in the --bacteria mode of Dorado polish for data basecalled with v5.2 models and improves the speed of 5mCG_5hmCG calling with v5.0 and v5.2 models, along with other enhancements and fixes.

• 2f7ed13 - Add v5.2 basecalling model support in --bacteria mode of Dorado polish
• 326be89 - Reduce CPU usage of 5mCG_5hmCG with v5.0 or v5.2 basecalling models
• 8b8a689 - Dorado polish and variant now fail on exceptions to avoid dropped data
• e6c1219 - Dorado polish and variant allow .fna and .fna.gz as input draft/reference
• e26e532 - Dorado polish and variant error message that data from duplex models are not supported
• c33c626 - Add [email protected] (duplicate of v1.3) to avoid Dorado download collision

v1.0.0

[1.0.0] (21 May 2025)

We are pleased to announce the release of Dorado v1.0, delivering new models and capabilities for Oxford Nanopore data analysis. This release introduces:

• New v5.2 basecalling model support for DNA and RNA, including:
  • More accurate HAC basecalling
  • Increased canonical basecalling accuracy in DNA with high 6mA density
  • New 2'Ome modification models for RNA: 2OmeG, m5C_2OmeC, inosine_m6A_2OmeA, and pseU_2OmeU
  • More accurate DNA modification models, including a significant reduction in false positive rates for 6mA
  • Significant speed enhancements to all-context modification models in both DNA and RNA, when using NVIDIA GPUs
• A preview release of the new Dorado variant command for small variant calling
  • This is currently only available with v5.0 HAC basecalls. Compatibility with v5.2 models will be coming soon.
• Poly(A) updates, including:
  • The new pa:B:i tag, described in the Dorado SAM spec
• Other fixes and improvements, including updates to logging and improved error messages

This release also removes support for Fast5 files and basecalling models for DNA R10.4.1 4kHz data, DNA R9.4.1, and RNA002. Support for these conditions will continue to be available with Dorado 0.9.6 and Fast5 files can be converted to POD5 using our online conversion tool or the POD5 Python tools.

• 90db41a - New v5.2.0 basecalling models for DNA and RNA
• 63bfbad - Preview release of new Dorado variant command for small variant calling
• 4df8bfe - Improve speed of basecalling with modified bases
• f0a82e3 - Add pa:B:i tag with signal positions related to the poly(A/T) estimation
• e1414d6 - Add poly(A/T) tail length to Dorado summary file
• cd7998e - Add beta support for RNA barcoding kit
• 746a172 - Remove support for fast5 files
• e15a69b - Remove support for DNA R10.4.1 4kHz basecalling models
• 8b7ef0f - Remove support for DNA R9.4.1 and RNA002 basecalling models
• d689102 - Add [info] log line showing the mapping of CUDA device index to device name
• af0e4fd - Improve error message when --resume-from file is missing CL header
• 8dbcc10 - Tweak logging, including adding a success message from Dorado downloader
• 0afabf3 - Reduce Transformer and modified base batch sizes when using Apple Silicon
• d3dcc83 - Build Dorado as C++20
• c63060e - Move default build to Rocky 8 to support Linux systems using glibc >= 2.28
• 967b117 - Use higher performance FP8 code path for Transformer models on C.C 8.9, 9.0, 10.0 and 12.x

v0.9.6

[0.9.6] (16 Apr 2025)

This release of Dorado introduces fixes and improvements following the Dorado 0.9.5 release, including a fix to support v5 SUP basecalling models on Turing (Compute Capability 7.5) GPUs and a fix to correct mapQ values reported by Dorado aligner.

• 4ea6018 - Update to torch 2.6.0.2 for CUDA Linux x64 and Windows to add support for Turing GPUs
• e24a7d0 - Modest speed improvement of Transformer based SUP basecalling models on Volta GPUs
• 74f9bdd - Improve handling of exceptions in Dorado polish
• a1ebe19 - Add Cuda 12.8 Rocky Linux 8 builds
• f34d0b8 - Fix mapQ values reported by Dorado aligner

v0.9.2

[0.9.2] (16 Apr 2025)

This release of Dorado matches the version included in MinKNOW 25.03 and Dorado Basecall Server 7.8.3. This release is only recommended to users who require matching performance to MinKNOW 25.03. For all other users, we recommend the latest version of Dorado (≥0.9.5) to benefit from all available improvements. Changes from 0.9.1 in this release include:

• 48b513d - Fix crashes when running Dorado basecaller on systems with heterogeneous GPUs
• 8a67020 - Improve speed of VCF output in Dorado polish
• c7de7e4 - Bump third party library versions

v0.9.5

[0.9.5] (31 March 2025)

This release of Dorado delivers significant performance improvements and broader hardware support:

• Major speed increase of the v5 SUP basecalling models on Volta (7.0) and Orin (8.7) GPUs
• Enhanced speedups for HAC basecalling models on Nvidia Hopper GPUs (Compute Capability 9.0)
• Support for Nvidia Blackwell GPUs (Compute Capabilities 10.0 and 12.0)

Additional updates include improvements to primer sequence trimming, read orientation inference (if inferred, this will be added to the TS tag in SAM/BAM) and UMI tag sequence detection (if found, these are included in the BAM/SAM output using the RX tag). Poly(A) tail estimation accuracy has been improved, along with the stability and speed of Dorado polish and correct. Dorado aligner now supports split indexes, and the internal dependency on minimap2 has been updated to v2.28.

Note: Dorado has moved to a building with CUDA 12.8, which increases the minimum supported Nvidia driver version to 525.105. The minimum supported macOS version is now 13.0 (Ventura). For Orin platforms, JetPack 6.2 (based on Ubuntu 22.04 'Jammy Jellyfish') is now required.

• 833a1aa - Improve speed of Transformer based SUP basecalling models on Volta GPUs
• d46286e - Improve speed of LSTM HAC basecalling models on Hopper GPUs
• c1763b6 - Enable Transformer SUP basecalling model performance improvements on Orin
• ee7b7cc - Add caching of automatic batch size for H100 GPUs
• c946a46 - Update minimum macOS to Ventura (version 13.0)
• 99c3cb2 - Remove support for Focal and CUDA 11 on Orin
• bffa71d - Update to JetPack 6.2 for Orin
• 78260d4 - Update to torch v2.6.0
• 0407778 - Update to minimap2 v2.28
• 7f4c575 - Add cDNA preprocessing in Dorado basecaller and trim, including UMI tag and strand orientation detection
• 3232e5b - Poly(A) tail estimation accuracy improvements
• 1d384dc - Poly(A) tail estimation now consistently emits pt:i BAM tag
• 56bef24 - Ensure all valid poly(A) tails in plasmids are captured during estimation
• 517d611 - Improve primer detection for kits with barcodes
• 04915f3 - Fix trimming of RNA reads affected by mux change and unblocking
• 598d17c - Add Dorado aligner support for split indexes
• c973914 - Fix Dorado aligner output non-determinism
• b02d3a7 - Dorado polish speed improvements
• 49d5aa8 - Prevent overlapping variants in Dorado polish VCF output
• e0eb79b - Dorado correct speed improvements in inference stage
• d520323 - Ensure Dorado does not take GPU memory on unspecified devices when using -x or --device
• 48b513d - Fix crashes when running Dorado basecaller on systems with heterogeneous GPUs
• e8a3217 - Fix inconsistency in calculation of sm/sd BAM tag values
• 12ffaa3 - Improve error handling in model download
• 4984a53 - Dorado trim logging now reports number of reads trimmed
• 8557882 - Change short adapter and/or primer trimming warning to trace message
• 990acb5 - Add note in README for getting good performance from GPUs on Windows
• 9d83b96 - Add sha256 checksums to all downloads
• 4661a77 - Fix custom sequence format documentation
• 6ed2865 - Drop macOS/x64 pipelines

v0.9.1

[0.9.1] (21 Jan 2025)

This release of Dorado brings significant basecalling speed improvements for Nvidia GPUs with compute capabilities 8.6 (Ampere – e.g., RTX A6000), 8.7 (Ampere – e.g., Orin family), and 8.9 (Ada Lovelace). Additionally, dorado polish receives major enhancements, including the introduction of the --bacteria flag, which optimizes basecalling for native bacterial and methylated DNA. The updated dorado polish is now compatible with data basecalled using v4.3 and v4.2 models and serves as a beta-stage replacement for Medaka.

• d46286e - Upgraded to Koi v0.5.3 for optimised basecalling on CC 8.6, 8.7 & 8.9
• 9c83c49 - Added VCF output to the Dorado polish
• d859977 - Improve messaging when an unrecognised model complex is specified
• 06f6239 - Support for legacy count models in Dorado polish
• e51d64f - Fixed precision issue with Dorado polish qscore normalisation
• d50494d - Added --bacteria support for Dorado polish
• f3fa447 - Added ULK114 kits to adapter/primer trimming LUT
• 1a4316e - Dorado polish algorithm improvements and refactoring
• a0fc960 - Dorado polish option -o now writes to a directory
• 7ae31a8 - Dorado polish fix for parsing FASTA file inputs
• b382344 - Dorado polish models are now included in Dorado download --list, --list-yaml, and --list-structured
• adebcc5 - Ensure Q-score and read length filters are applied to reads after trimming
• 96d574d - Enforce use of --kit-name for Dorado demux custom barcodes to prevent unclassified reads
• 62c86d0 - Support CUDA_VISIBLE_DEVICES for UUIDs or integers
• c39f906 - Prevent crashing in auto batchsize due to short chunksizes

v0.9.0

[0.9.0] (16 Dec 2024)

This major release of Dorado introduces several new features and enhancements. The polish command, currently experimental, is optimised for refining draft assemblies of human genomes. This release also adds faster DNA modification calling models and improved 6mA false positive rate (FPR) in native human samples. Barcode demultiplexing accuracy has been significantly enhanced for kits with barcodes at both ends, including SQK-NBD114. Note that using custom barcode kits now requires the --kit-name option. A feature has been added to enable running dorado correct in blocks, allowing work to be divided into smaller pieces for easy submission to a compute cluster. Additional updates include the qs tag for mean basecall Q-scores in FASTQ output, an upgrade to POD5 to support systems with large page sizes, improvements to Poly(A) tail length estimation, and various bug fixes to enhance stability and functionality.

• 2b96c0b - New Dorado polish feature for assembly polishing
• 0bab166 - Faster modified base models for DNA 4mC_5mC, 5mC_5hmC, 5mCG_5hmCG, and 6mA
• e637166 - Enable running dorado correct in blocks, for easy submission to a compute cluster
• 40296da - Reduced false positive classification rates for kits with barcodes at both ends
• 35da003 - Improve barcode classification when barcodes can be on either end
• cbcdf38 - Only classify barcodes which are present on sample sheet if provided
• 2449d03 - Correct AF02F_14 and AH10R_80 barcodes from TWIST-96A-UDI
• 631e94c - Prevent Dorado demux from stripping alignment information when --no-trim is specified
• affea85 - Prevent missing filenames when using --emit-summary with Dorado demux
• 3dec15a - Improve poly(A) tail estimation accuracy, including with interrupted tails
• df57d34 - Limit poly(A) estimation to reads with plausible signal to prevent stalls in calculation
• 6cf701a - Add min_primer_separation option to custom poly(A) configuration
• bf51bd4 - Add qs tag with mean basecall Q-score to FASTQ output
• dac076d - Upgrade to POD5 v0.3.23 to support systems with large page sizes for POD5 and .fast5
• c7a7a58 - Prevent silent failure or segfault on Windows with bad custom barcode files
• 1e829d5 - Do not allow basecalling if target directory includes both POD5 and .fast5 files
• 05d0981 - Fix modified base trim for reverse-aligned BAM records
• afdb068 - Fix invalid MM tag after trimming when no mods are present
• 0d788d7 - Prevent crash when insufficient permissions to read an input file/folder
• dbece01 - Update custom barcoding documentation to accurately reflect demultiplexing logic
• 6db40ec - Correct model context info shown in dorado download --list-structured
• 03acc12 - Use the -o short option only for --output-dir and not for --overlap
• 8d9c017 - Added support for reading gzipped compressed FASTQ files

v0.7.4

[0.7.4] (11 Dec 2024)

This release of Dorado matches the version included in MinKNOW 24.11 and Dorado Basecall Server 7.6.7. We only recommend this release to users who require matching performance to MinKNOW 24.11. Other users should use the latest version of Dorado (≥0.8.3) to benefit from all available improvements. Changes from 0.7.3 in this release include:

• 40296da - Reduced false positive classifications for NBD barcode kits
• 687f234 - Improved GPU batch submission timeout logic to improve basecall performance with low data throughput
• 03e5eec - Fix incorrect basecalls being emitted from SUP models on Apple Silicon
• 9a81cbc - Fix errors when running dorado basecaller on multi-GPU systems
• 3cc4de3 - Prevent "Too many open files" error when using --sort-bam with dorado demux
• f35c8cc - Fix bug when downloading models for dorado correct
• 4a28d58 - Always trim DNA adapter signal before processing RNA reads
• adc60ba - Package libcupti.so into ARM Linux builds
• c65b3fb - Added EXP-NBD114-24 alias for SQK-NBD114-24
• 69cb260 - Fix bug causing intermittent crashing with v5 SUP models
• a674dad - Update --help documentation for basecaller, duplex, and correct
• 6ec77c8 - Fix errors when performing duplex calling with modified bases
• cb6eee1 - Decouple alignment and inference stages in dorado correct
• 762e886 - Cache batch sizes to significantly reduce basecaller startup time on supported GPUs
• 9e5db84 - Fix duplicated alignment tags in re-aligned files
• 966c2ca - Update POD5 version to v0.3.15
• e9281fa - Emit an error message if header from input HTS file cannot be read
• fcb9d53 - Include run name in output files from dorado demux even if input files are FASTQ
• 7f42b8f - Warn and exit instead of crashing if a model path does not exist
• 7d74246 - Improve index file error handling
• 022901e - Fix JSON output when using --list-structured with dorado download

v0.8.3

[0.8.3] (11 Nov 2024)

This release of Dorado adds fixes and improvements to the Dorado 0.8.2 release, including a fix to SUP basecalling on Apple Silicon.

• 03e5eec - Fix incorrect basecalls being emitted from SUP models on Apple Silicon
• 987a237 - Improve error messaging when unable to run basecalling kernels on Apple GPU