Release of DeepVariant and DeepSomatic v1.9
DV: Now train on HG002 T2T-Q100. Error reduction of 12% for Illumina and 30% for PacBio on this truth set. 25% faster. DeepTrio is 5x faster (20h -> 4h).
DS: New models FFPE_TUMOR_ONLY for {WGS, WES}. Much improved WGS models.
github.com/google/deepv...
DV: Now train on HG002 T2T-Q100. Error reduction of 12% for Illumina and 30% for PacBio on this truth set. 25% faster. DeepTrio is 5x faster (20h -> 4h).
DS: New models FFPE_TUMOR_ONLY for {WGS, WES}. Much improved WGS models.
github.com/google/deepv...
Release DeepVariant 1.9.0 · google/deepvariant
DeepVariant:
In this version we have updated our training scheme for the HG002 sample with the newly released HG002-T2T truth set which improves accuracy against that truth set.
Our labeling metho...
github.com
May 13, 2025 at 8:23 PM
Release of DeepVariant and DeepSomatic v1.9
DV: Now train on HG002 T2T-Q100. Error reduction of 12% for Illumina and 30% for PacBio on this truth set. 25% faster. DeepTrio is 5x faster (20h -> 4h).
DS: New models FFPE_TUMOR_ONLY for {WGS, WES}. Much improved WGS models.
github.com/google/deepv...
DV: Now train on HG002 T2T-Q100. Error reduction of 12% for Illumina and 30% for PacBio on this truth set. 25% faster. DeepTrio is 5x faster (20h -> 4h).
DS: New models FFPE_TUMOR_ONLY for {WGS, WES}. Much improved WGS models.
github.com/google/deepv...
Release of DeepVariant v1.10
Phased VCF output for long-reads
Accuracy improvements for multi-allelic variants
Pangenome accuracy improvements (18% fewer errors)
Most technologies ~10% faster
RNA-seq is a full supported mode
DeepSomatic is 12-40% faster
github.com/google/deepv...
Phased VCF output for long-reads
Accuracy improvements for multi-allelic variants
Pangenome accuracy improvements (18% fewer errors)
Most technologies ~10% faster
RNA-seq is a full supported mode
DeepSomatic is 12-40% faster
github.com/google/deepv...
Release DeepVariant 1.10.0 · google/deepvariant
DeepVariant:
Continuous phasing: Long-read variant calls (PacBio and ONT) are now natively phased and phased output is generated for both vcf and gvcf formats.
Fuzzy channels: Added “fuzzy channel...
github.com
March 7, 2026 at 12:26 AM
Release of DeepVariant v1.10
Phased VCF output for long-reads
Accuracy improvements for multi-allelic variants
Pangenome accuracy improvements (18% fewer errors)
Most technologies ~10% faster
RNA-seq is a full supported mode
DeepSomatic is 12-40% faster
github.com/google/deepv...
Phased VCF output for long-reads
Accuracy improvements for multi-allelic variants
Pangenome accuracy improvements (18% fewer errors)
Most technologies ~10% faster
RNA-seq is a full supported mode
DeepSomatic is 12-40% faster
github.com/google/deepv...
The new Google DeepVariant now works for @pacbio.bsky.social Revio SPRQ chemistry data *and* long-read RNA-Seq (Kinnex / Iso-Seq) data! Go give it a try! 🧬🧪
github.com/google/deepv...
github.com/google/deepv...
Release DeepVariant 1.8.0 · google/deepvariant
In this release:
Small model integration: Speed increased by ~1.7x (40% runtime reduction) for WGS, PacBio, and ONT by introduction of additional small model. The small model identifies easy-to-...
github.com
December 6, 2024 at 8:45 PM
The new Google DeepVariant now works for @pacbio.bsky.social Revio SPRQ chemistry data *and* long-read RNA-Seq (Kinnex / Iso-Seq) data! Go give it a try! 🧬🧪
github.com/google/deepv...
github.com/google/deepv...
Pangenome-aware DeepVariant https://www.biorxiv.org/content/10.1101/2025.06.05.657102v1
June 6, 2025 at 10:48 PM
Pangenome-aware DeepVariant https://www.biorxiv.org/content/10.1101/2025.06.05.657102v1
Release of DeepVariant 1.8. Large speed improvement (~67% faster) via small model for easy sites. New Pangenome-aware option. Reduces error by ~30% for vg-mapped WGS, ~10% for BWA WGS, ~5% BWA exome. New config for custom model users, see release notes.
(github.com/google/deepv...)
(github.com/google/deepv...)
December 5, 2024 at 5:57 PM
Release of DeepVariant 1.8. Large speed improvement (~67% faster) via small model for easy sites. New Pangenome-aware option. Reduces error by ~30% for vg-mapped WGS, ~10% for BWA WGS, ~5% BWA exome. New config for custom model users, see release notes.
(github.com/google/deepv...)
(github.com/google/deepv...)
Ok so this #deepvariant caller update from #google is BIG.
github.com/google/deepv...
- 1.6x time reduction (40% increase ??)
- Pangenome integration
- New model and improvements over previously model
Excited to try this new update 🧬💻
#bioinformatics #wgs #ngs #variantcalling
github.com/google/deepv...
- 1.6x time reduction (40% increase ??)
- Pangenome integration
- New model and improvements over previously model
Excited to try this new update 🧬💻
#bioinformatics #wgs #ngs #variantcalling
Release DeepVariant 1.8.0 · google/deepvariant
In this release:
Small model integration: Speed increased by ~1.7x (40% runtime reduction) for WGS, PacBio, and ONT by introduction of additional small model. The small model identifies easy-to-...
github.com
December 9, 2024 at 6:39 AM
Ok so this #deepvariant caller update from #google is BIG.
github.com/google/deepv...
- 1.6x time reduction (40% increase ??)
- Pangenome integration
- New model and improvements over previously model
Excited to try this new update 🧬💻
#bioinformatics #wgs #ngs #variantcalling
github.com/google/deepv...
- 1.6x time reduction (40% increase ??)
- Pangenome integration
- New model and improvements over previously model
Excited to try this new update 🧬💻
#bioinformatics #wgs #ngs #variantcalling
Pangenome-aware DeepVariant www.biorxiv.org/content/10.1... 🧬🖥️🧪 github.com/google/deepv...
Pangenome-aware DeepVariant
Population-scale genomics information provides valuable prior knowledge for various genomic analyses, especially variant calling. A notable example of such application is the human pangenome…
www.biorxiv.org
June 9, 2025 at 6:30 PM
Pangenome-aware DeepVariant www.biorxiv.org/content/10.1... 🧬🖥️🧪 github.com/google/deepv...
Release of DeepVariant v1.6.
Support for haploid regions, chrX/Y.
Workflow for Pangenome FASTQ-to-VCF.
Major DeepTrio improvements for de novo variants.
Models for CompleteGenomics T7, G400
Add NovaSeqX to training data
Release by Kishwar Shafin
github.com/google/deepv...
Support for haploid regions, chrX/Y.
Workflow for Pangenome FASTQ-to-VCF.
Major DeepTrio improvements for de novo variants.
Models for CompleteGenomics T7, G400
Add NovaSeqX to training data
Release by Kishwar Shafin
github.com/google/deepv...
Release DeepVariant 1.6.0 · google/deepvariant
Improved support for haploid regions, chrX and chY. Users can specify haploid regions with a flag. Updated case studies show usage and metrics.
Added pangenome workflow (FASTQ-to-VCF mapping with V...
github.com
October 26, 2023 at 4:32 PM
Release of DeepVariant v1.6.
Support for haploid regions, chrX/Y.
Workflow for Pangenome FASTQ-to-VCF.
Major DeepTrio improvements for de novo variants.
Models for CompleteGenomics T7, G400
Add NovaSeqX to training data
Release by Kishwar Shafin
github.com/google/deepv...
Support for haploid regions, chrX/Y.
Workflow for Pangenome FASTQ-to-VCF.
Major DeepTrio improvements for de novo variants.
Models for CompleteGenomics T7, G400
Add NovaSeqX to training data
Release by Kishwar Shafin
github.com/google/deepv...
Anyone in my Sky-ome have experience with pangenome-based population genomics? Calling SNPs with Deepvariant is giving me a headache 😭
December 13, 2024 at 11:43 AM
Anyone in my Sky-ome have experience with pangenome-based population genomics? Calling SNPs with Deepvariant is giving me a headache 😭
Pangenome-aware DeepVariant [new]
Leverages pangenomes for variant calling by creating pileup images of reads and haplotypes, using a CNN to improve genotype inference.
Leverages pangenomes for variant calling by creating pileup images of reads and haplotypes, using a CNN to improve genotype inference.
June 6, 2025 at 11:02 PM
Pangenome-aware DeepVariant [new]
Leverages pangenomes for variant calling by creating pileup images of reads and haplotypes, using a CNN to improve genotype inference.
Leverages pangenomes for variant calling by creating pileup images of reads and haplotypes, using a CNN to improve genotype inference.
#PacBio long-read data was used to build a new genome benchmark that reduced DeepVariant errors by 34 percent in hard-to-analyze regions. Platinum Pedigree improves variant calling accuracy and sets a new bar for how AI is trained and validated in genomics.
Press release here: bit.ly/4mqcBKm
Press release here: bit.ly/4mqcBKm
August 4, 2025 at 3:04 PM
#PacBio long-read data was used to build a new genome benchmark that reduced DeepVariant errors by 34 percent in hard-to-analyze regions. Platinum Pedigree improves variant calling accuracy and sets a new bar for how AI is trained and validated in genomics.
Press release here: bit.ly/4mqcBKm
Press release here: bit.ly/4mqcBKm
New vg v1.74.0 "Petrie" update includes:
🔹Single-command way to run the haplotype sampling mapping workflow with vg giraffe
🔹 Minimap2-inspired alignment scoring approach for giraffe
🔹Ability to write v.1 GBZ files for compatibility with pangenome-aware DeepVariant
github.com/vgteam/vg/re...
🔹Single-command way to run the haplotype sampling mapping workflow with vg giraffe
🔹 Minimap2-inspired alignment scoring approach for giraffe
🔹Ability to write v.1 GBZ files for compatibility with pangenome-aware DeepVariant
github.com/vgteam/vg/re...
May 12, 2026 at 7:26 PM
New vg v1.74.0 "Petrie" update includes:
🔹Single-command way to run the haplotype sampling mapping workflow with vg giraffe
🔹 Minimap2-inspired alignment scoring approach for giraffe
🔹Ability to write v.1 GBZ files for compatibility with pangenome-aware DeepVariant
github.com/vgteam/vg/re...
🔹Single-command way to run the haplotype sampling mapping workflow with vg giraffe
🔹 Minimap2-inspired alignment scoring approach for giraffe
🔹Ability to write v.1 GBZ files for compatibility with pangenome-aware DeepVariant
github.com/vgteam/vg/re...
Added SPRQ to PacBio training, reducing Indel error on SPRQ by 26%. Added Platinum Pedigree training data for PacBio model, reducing errors by 34% on more extensive Platinum truth. New model and case study for Kinnex/Mas-Seq/Iso-Seq. Additional speed options for GPU pipelines 2/3
December 5, 2024 at 5:57 PM
Added SPRQ to PacBio training, reducing Indel error on SPRQ by 26%. Added Platinum Pedigree training data for PacBio model, reducing errors by 34% on more extensive Platinum truth. New model and case study for Kinnex/Mas-Seq/Iso-Seq. Additional speed options for GPU pipelines 2/3
Yeah, CNN-based algorithms are transforming pathology for the better, deepvariant is quite good, alphafold is revolutionary, etc.
But those are specialized models, not these LLMs that are taking over everywhere else
But those are specialized models, not these LLMs that are taking over everywhere else
October 17, 2025 at 2:29 PM
Yeah, CNN-based algorithms are transforming pathology for the better, deepvariant is quite good, alphafold is revolutionary, etc.
But those are specialized models, not these LLMs that are taking over everywhere else
But those are specialized models, not these LLMs that are taking over everywhere else
8/ DeepVariant (Google): AI-powered variant calling from NGS data. Uses deep learning for high accuracy.
Open source and integrates with Google Cloud for scalability. github.com/google/deep...
Open source and integrates with Google Cloud for scalability. github.com/google/deep...
GitHub - google/deepvariant: DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data.
DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data. - google/deepvariant
github.com
February 18, 2026 at 2:15 PM
8/ DeepVariant (Google): AI-powered variant calling from NGS data. Uses deep learning for high accuracy.
Open source and integrates with Google Cloud for scalability. github.com/google/deep...
Open source and integrates with Google Cloud for scalability. github.com/google/deep...
Existing long-read SNP callers (DeepVariant, longshot...) have been developed for diploid genomes. Deep-learning methods are trained on [human] genomic data. Statistical methods contain assumption that do not hold for metagenomics.
December 4, 2025 at 1:18 PM
Existing long-read SNP callers (DeepVariant, longshot...) have been developed for diploid genomes. Deep-learning methods are trained on [human] genomic data. Statistical methods contain assumption that do not hold for metagenomics.
In Arabidopsis, DeepVariant achieved good balance with FP / FN when we use AnchorWave alignments as ground truth. It’s more conservative with much less FP ( much less false heterozygous sites than GATK and much faster)
April 4, 2026 at 6:36 PM
In Arabidopsis, DeepVariant achieved good balance with FP / FN when we use AnchorWave alignments as ground truth. It’s more conservative with much less FP ( much less false heterozygous sites than GATK and much faster)
Presenting interesting benchmark datasets from Google DeepVariant case studies (haven't seen these before, they look useful): github.com/google/deepv...
GitHub - google/deepvariant: DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data.
DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data. - google/deepvariant
github.com
July 16, 2024 at 2:38 PM
Presenting interesting benchmark datasets from Google DeepVariant case studies (haven't seen these before, they look useful): github.com/google/deepv...
Why does GATK exist when bcftools does a perfectly good job, especially given that they both end up feeding DeepVariant or DeepSomatic? Why isn’t everything shoved into Parquet via GA4GH APIs?
You get the idea
You get the idea
November 7, 2025 at 12:45 AM
Why does GATK exist when bcftools does a perfectly good job, especially given that they both end up feeding DeepVariant or DeepSomatic? Why isn’t everything shoved into Parquet via GA4GH APIs?
You get the idea
You get the idea
1/
What reference genome should you use?
Sounds easy. It’s not.
GRCh37? GRCh38? hs37d5?
Have you heard of T2T or the new pan-genome-aware DeepVariant?
This matters more than you think.
www.biorxiv.org/content/10....
What reference genome should you use?
Sounds easy. It’s not.
GRCh37? GRCh38? hs37d5?
Have you heard of T2T or the new pan-genome-aware DeepVariant?
This matters more than you think.
www.biorxiv.org/content/10....
June 10, 2025 at 1:15 PM
1/
What reference genome should you use?
Sounds easy. It’s not.
GRCh37? GRCh38? hs37d5?
Have you heard of T2T or the new pan-genome-aware DeepVariant?
This matters more than you think.
www.biorxiv.org/content/10....
What reference genome should you use?
Sounds easy. It’s not.
GRCh37? GRCh38? hs37d5?
Have you heard of T2T or the new pan-genome-aware DeepVariant?
This matters more than you think.
www.biorxiv.org/content/10....
Direct paper link:
"Deep learning approaches, specifically Clair3 and DeepVariant, deliver high accuracy in SNP and indel calls from ONT data, outperforming Illumina-based methods, with Clair3 achieving median F1 scores of 99.99% for SNPs and 99.53% for indels." #pp
doi.org/10.7554/eLif...
"Deep learning approaches, specifically Clair3 and DeepVariant, deliver high accuracy in SNP and indel calls from ONT data, outperforming Illumina-based methods, with Clair3 achieving median F1 scores of 99.99% for SNPs and 99.53% for indels." #pp
doi.org/10.7554/eLif...
Benchmarking reveals superiority of deep learning variant callers on bacterial nanopore sequence data
Nanopore sequencing effectively resolves key challenges in bacterial small variant detection, providing a more accurate alternative to traditional sequencing methods.
doi.org
December 3, 2024 at 8:18 PM
Direct paper link:
"Deep learning approaches, specifically Clair3 and DeepVariant, deliver high accuracy in SNP and indel calls from ONT data, outperforming Illumina-based methods, with Clair3 achieving median F1 scores of 99.99% for SNPs and 99.53% for indels." #pp
doi.org/10.7554/eLif...
"Deep learning approaches, specifically Clair3 and DeepVariant, deliver high accuracy in SNP and indel calls from ONT data, outperforming Illumina-based methods, with Clair3 achieving median F1 scores of 99.99% for SNPs and 99.53% for indels." #pp
doi.org/10.7554/eLif...
Note: OLD POST! (2023), but I just noticed it.
While it's nice to see comparisons, why compare an (at the time) 2 year old GATK against a 5 year old bcftools?
Since then both have come on a lot. It'd be interesting to see new independent comparisons. (Neither can hold up to deepvariant now.)
While it's nice to see comparisons, why compare an (at the time) 2 year old GATK against a 5 year old bcftools?
Since then both have come on a lot. It'd be interesting to see new independent comparisons. (Neither can hold up to deepvariant now.)
Important comparison of Bcftools and GTK in simulated Drosophila genomes: "by benchmark analyses with a simulated insect population...Bcftools mpileup performs better than GATK HaplotypeCaller in terms of recovery rate and accuracy regardless of mapping software."
The evaluation of Bcftools mpileup and GATK HaplotypeCaller for variant calling in non-human species...
Scientific Reports - The evaluation of Bcftools mpileup and GATK HaplotypeCaller for variant calling in non-human species
www.nature.com
September 18, 2025 at 8:58 PM
Note: OLD POST! (2023), but I just noticed it.
While it's nice to see comparisons, why compare an (at the time) 2 year old GATK against a 5 year old bcftools?
Since then both have come on a lot. It'd be interesting to see new independent comparisons. (Neither can hold up to deepvariant now.)
While it's nice to see comparisons, why compare an (at the time) 2 year old GATK against a 5 year old bcftools?
Since then both have come on a lot. It'd be interesting to see new independent comparisons. (Neither can hold up to deepvariant now.)
5 AI tools changing biology NOW:
AlphaFold: Solves protein folding
IBM Watson: Analyzes complex bio-data
Nvidia Clara: Powers genomics & imaging
DeepVariant: IDs genetic variants
BenchSci: Plans lab experiments
It isn't fake; it's here. Did I miss any?
#AI #Biology #Zoology #BioTech #Science
AlphaFold: Solves protein folding
IBM Watson: Analyzes complex bio-data
Nvidia Clara: Powers genomics & imaging
DeepVariant: IDs genetic variants
BenchSci: Plans lab experiments
It isn't fake; it's here. Did I miss any?
#AI #Biology #Zoology #BioTech #Science
TOP 5 AI Tools Every Biologist Must Know! #ai #biology #top
YouTube video by Biotecnika
youtu.be
November 6, 2025 at 2:33 AM
Release led by DeepVariant tech lead Kishwar Shafin. Team Engineering manager Pi-Chuan Chang. Small model work led by Lucas Brambrink. Pangenome-aware led by Mobin Asri and Juan Carlos Mier. Fast pipeline by Alexey Kolesnikov. Kinnex/MAS-Seq model by Daniel Cook and Shiyi Yin from Verily. 3/3
December 5, 2024 at 5:57 PM
Release led by DeepVariant tech lead Kishwar Shafin. Team Engineering manager Pi-Chuan Chang. Small model work led by Lucas Brambrink. Pangenome-aware led by Mobin Asri and Juan Carlos Mier. Fast pipeline by Alexey Kolesnikov. Kinnex/MAS-Seq model by Daniel Cook and Shiyi Yin from Verily. 3/3