Feature additions:

• Addition of modified base models for v5.0 basecalling models
  • All models (except for inosine) provided with SUP and HAC support
  • DNA
    • 5mC+5hmC, 5mC+4mC, 6mA
  • RNA (all models now all-contexts)
    • m6A, Pseudouridine, Inosine
• Add remora model inspect command
• Add remora dataset copy command
• Allow multiple models to be used with remora infer from_pod5_and_bam (support C-mod and A-mod calling simultaneously)
• Update learning rate scheduler to use cosine decay
• Support IUPAC revcomp
• Picoamp scaling enabled (model maintain use of scaling via k-mer levels)
• Allow signal padding for chunks at the end of a read
• Add parameter to limit the size of a dataset merge
• Allow training and validation from core or config dataset
• Reference-anchored inference documentation

Bug fixes:

• Extend access to missing_metrics_ok to allow more robust use of API for read loading
• Pin plotnine version
• Fix bug for chunk extraction over entire read
• Fix bug in m6A model specification

Add v4.3 modified base models

• 5mC+5hmC CG-context SUP
• 5mC+5hmC CG-context HAC
• 5mC+5hmC all-context HAC and SUP
• 6mA all-context HAC and SUP
• 4mC+5mC all-context SUP (research model)

Bug fixes:

• Fix overflow on re-squiggle (#138)
• Fix issue with split reads in API (#134)

This version adds several new features as well as general bug fixes and optimizations.

Key Improvements:

• A major Remora datasets update
  • Easier dataset composition and manipulation
    • Flexible dataset mixing, allowing use of randomers, native, enzymatic, PCR, spike-in, and other dataset types
    • Datasets defined by configuration file, which can be generated automatically
  • Larger datasets enabled
    • Model training has now been demonstrated on over one billion training chunks
  • Easier hyper-parameter tuning at training time
• Enhanced signal and metrics plotting and exploration interface
• Improved model inference speed
• Full RNA support, including an m6A model - also available for production modified base calling through Dorado
• ChEBI code support
  • Allow any modified base with full pipeline support
• Split reads support
• Use latest POD5 update
  • Allow single POD5 or directory of POD5 files as input
• Various bug fixes

Fix for cython 3.0 update

Minor bug fixes and run settings checks

• Fixes #82 and #83
• Added check to avoid original issue for #82

• Add all-context models to repository
• Add finetune option to model training

This version adds several new features as well as general bug fixes and optimizations.

Major features:

• Raw signal plotting allows users to visualize raw Nanopore signal aligned to a reference
  • In addition, the Remora API supports efficient and easy-to-use access to pre-read per-site signal based statistics (such as notebooks in repository).
  • Allows users to explore modified bases in signal space to gain intuition for modified base model training.
  • This replaces the Tombo signal visualization and metrics extraction features.
• Infer performance optimization and bug fix on certain systems
  • Note that Dorado is still preferred as the production modified basecalling platform, but these improvements allow users to perform modified basecalling after canonical basecalling more efficiently.
• Initial RNA Support
  • Support 3' to 5' signal in Remora datasets and models
• Various training improvements
  • Batch balancing
  • Dynamic training data filtering

Remora v2.0.0 release

Feature additions:

• Updated kit14 5mC+5hmC models
• Simplified POD5+BAM input pipeline
• Remove ONNX model format (pytorch only unified with Dorado)
• Automatic model downloads
• Inference and validation from modBAM format
• Duplex modified base calling
• Remore Taiyaki/Megalodon dependency
• Basecall-anchored training

Remora v1.1.1 release

Feature additions:

• Guppy-compatible model export including version 1 Remora models

Bug Fixes

• onnxruntime protobuf dependency version issue
• remora validate from_modbams using strand from --regions-bed
• Fix big in unused chunk extraction code

Remora 1.1.0 release

Feature additions:

• Kit14 (R10.4.1 E8.2) model releases
• Kit12 (R10.4 E8.1) model updates
• Improved validation tools (remora validate from_modbams)
• Update to new modbam tag specifications (? notation)
• Improved support for custom training data
• Fix scaling bug causing numpy divide by zero warning
• All-context 5mC Kit14 research model added to Rerio (https://github.com/nanoporetech/rerio#remora-models)