#Basecalling
1/6 New preprint & work in progress: Brieno, a model of raw nanopore signal for many tasks. A self-supervised backbone, with a task-specific head, classifies species, demultiplexes barcodes, basecalls, and detects 5mC and 6mA methylation.

doi.org/10.64898/202...
November 6, 2026 at 2:02 PM
I'm booting up a new gaming laptop that instructional support bought so i can run ONT MinIONs w/ basecalling live in classes, and it's asking me to name the computer, so of course I named him Gru.
October 30, 2026 at 7:09 PM
2/6 Brieno uses a residual-VQ tokenizer to convert the raw current into discrete tokens. A Transformer is pretrained by masked-token prediction on 118 M R9.4.1 reads from 16 public datasets on CSC's Roihu supercomputer. Pretraining improves all four downstream tasks over the tokens alone.
November 6, 2026 at 2:02 PM
4/6 In basecalling, fine-tuned Brieno comes close to Dorado sup and exceeds Dorado hac on every test set. It outperforms Dorado's 5mC model at CpG sites and the NEMO 6mA caller at GATC sites.
November 6, 2026 at 2:02 PM
Check the reverse vs fwd reads. Systematic basecalling error can cause some funky things. Noticed this in repeat regions especially, and also with my poa-consensus tool (which you might like actually 😁)
October 30, 2026 at 9:54 AM
Yup directly when basecalling
October 13, 2026 at 8:31 AM
you mean when basecalling?
October 13, 2026 at 2:30 AM
The chemistry is the easy part now. Whether "2 hours" is real depends on what happens after the flow cell: basecalling on the right GPU, a locked and validated pipeline, the reference choice, signed-off QC thresholds, and a report a clinician can read in 30 seconds. 2/4
October 5, 2026 at 3:33 PM
Just looked at PromethION hardware requirements and cannot stop thinking about one thing - these requirements are extremely high (almost ridiculous) just because ONT pushes live data analysis & basecalling during the sequencing.
However, this is unnecessary & makes such a system way more expensive.
September 24, 2026 at 5:35 PM
There is alredy existing much more efficient way: local system is used only for data acquisition, while all data processing [basecalling, etc] performed on HPC.
This approach drastically reduces the requirements to the local system and actually may save enough money to hire a postdoc for 2-3 years.
September 24, 2026 at 5:44 PM
Power situation in the lab had been dicey lately - I think the already lackluster grid is overloaded by people sneaking in AC/coolers/

So every time checking nvtop on my workstation is a lot more thrilling than it should be. I won't be able to meet deadlines if another basecalling run cancels out.
September 10, 2026 at 12:35 PM
Functional chimeric mRNAs encode proteins in mammalian immunity
RNA libraries were prepared using the Direct RNA Sequencing kit (SQK-RNA002). Sequencing was performed on the Oxford Nanopore PromethION P24 sequencer using the FLO-PRO002 flow cell. Sequencing was performed using MinKNOW v.22.03.4 and MinKNOW Core v.5.0.0 with a pore scan frequency of 1.5 h, active channel selection set to on and reserved pores set to on. Libraries were run for 72 h. Basecalling was performed using Bream v.7.0.9 and Guppy v.6.0.7. Data processing was performed on an Ubuntu 20.04 system. Sequencing was performed by the BioMicro Center at Massachusetts Institute of Technology. Computational analysis of long-read direct RNA-seq data Individual passing FASTQ files from the same sample were merged to generate a single FASTQ file per sample. Except in the case of JAFFAL17 v.2.2 and v.2.3, all mouse data analyses were performed using reference files from GENCODE58 release M28 (GRCm39), and all human data analyses were performed using reference files from GENCODE Release 43 (GRCh38.p13). For JAFFAL, matched reference files were downloaded from the UCSC genome browser59 and prepared using the instructions provided on the JAFFAL wiki (https://github.com/Oshlack/JAFFA/wiki). Genomic alignments for NanoPlot60 v.1.42.0 were performed using Minimap261 v.2.24 with the parameters -ax splice -uf -k14 and aligned reads were processed...
www.nature.com
September 2, 2026 at 7:38 PM
🍻 tracy-genomics 🍻

Basecalling, alignment, assembly and deconvolution of Sanger chromatograms

🔗 https://www.gear-genomics.com/docs/tracy/

#homebrew #newpkg #macos #linux #formula
August 19, 2026 at 1:26 PM
How much more compute did basecalling take with the new model?
June 5, 2025 at 12:32 PM
Does anyone have updated benchmarks for #Dorado basecalling on different GPUs, including Apple Silicon? #Nanopore #Sequencing
December 26, 2024 at 10:38 AM
This is interesting. I had no idea you need gpu for Ont basecalling
April 17, 2026 at 1:19 AM
Who do we have to lobby to be able to submit pod5 files from @nanoporetech.com to SRA? The actual only way is to use guppy-4.2.2 basecalling fast5 files and that makes very little sense and it is very unpractical! #sequencing #ONT #SRA #NCBI #reproducibility #OpenScience
December 10, 2025 at 1:18 AM
We provide specific recommendations for sequencing depth, basecalling model and analysis methods depending on the resources available and clinical context of the user. Analysis time is negligible, the biggest time savings come from choosing your basecalling model and depth requirements /4
July 30, 2025 at 2:13 AM
Nanopore data has come a long way the last few years, and offers users a huge amount of flexibility in how they want to generate data. We took all combinations of sequencing chemistry, basecalling, depth & assembly status from 3 Gram-negative species groups and compared to hybrid gold standards /2
July 30, 2025 at 2:12 AM
Pleased to say that our preprint benchmarking Nanopore data for MLST, cgMLST, cgSNP & AMR typing from bacterial isolates is out! TL;DR you can get almost perfect results from 50x depth using live SUP basecalling with a GPU in under 20 hours #microsky#IDsky 🦠🧬🖥️ /1
www.medrxiv.org/content/10.1...
July 30, 2025 at 2:11 AM
Pawsey Advances Genomic Research with AMD GPU Support for Basecalling Feb. 25, 2025 — Research...

https://www.hpcwire.com/off-the-wire/pawsey-advances-genomic-research-with-amd-gpu-support-for-basecalling/

Event Attributes
February 26, 2025 at 2:53 PM
What custom-tuned basecalling model are you using? Something trained in-house?
Regarding whether Dorado polish would help or make things worse, the best check would be to do some Illumina sequencing as well.
July 8, 2026 at 5:09 AM
New blog post!
I analyse the new hac@v6.0.0 basecalling model from @nanoporetech.com and discuss the conspicuous lack of a new sup model:
rrwick.github.io/2026/06/11/d...
Dorado v2.0.0: no more sup?
a blog for miscellaneous bioinformatics stuff
rrwick.github.io
June 11, 2026 at 5:14 AM
Yes, I used 6mA,4mC_5mC when basecalling. I didn't see any 4mC calls in the methylated_sites.tsv file. Also couldn't see anything suspicious in the bedMethyl file.
The motif is often GAGCTC, but GAGCTA, GAGCTG and GAGCTT came up too 😕
May 28, 2025 at 1:18 AM