PengNi / deepsignal-plant

Detecting methylation using signal-level features from Nanopore sequencing reads of plants
GNU General Public License v3.0
57 stars 12 forks source link

updated pipeline relying on pod5 file format? #29

Open AlineMuyle opened 1 year ago

AlineMuyle commented 1 year ago

Dear Dr Peng, I was wondering if you are considering updating your deepsignal-plant pipeline with the new compressed pod5 format? Including the resquiggle step. That would be extremely useful because my fast5 files are very large. Thank you for your help. Best wishes

PengNi commented 1 year ago

Hi @AlineMuyle , Thank you very much for using deepsignal-plant. We are planning to re-design the pipeline, including changing fast5 to pod5/slow5. However, it may not be done in next few months. As you know, the Tombo package only accepts single-fast5 format, which is not easy to be used as an interface. I will release a new version once we complete.

Best, Peng

AlineMuyle commented 1 year ago

Thank you @PengNi I will keep an eye on your updates!

jcolicchio-soundag commented 1 year ago

We are also looking forward to being able to use deepsignal with pod5s, and also with having a trained model from the new R10.4 flow cells! Excited to hear how progress on this is going.

AlineMuyle commented 1 year ago

Hi @PengNi I was wondering if you have any news on the updated pipeline with pod5/slow5? My datasets are very large and it would be super useful. Thank you for your help. Best wishes

PengNi commented 1 year ago

@AlineMuyle , sorry for the delay and thanks very much for your patience, we now have sequenced some new data in pod5 format, we are planning to release a new version before the end of this year, including for R10.4.1 flowcell.

Best, Peng

drfultz commented 10 months ago

@PengNi Any updates on a newer version of this tool that doesn't need tombo / works with pod5 and R10 pore data? Our group uses your tool regularly (thanks for a nice tool!), but we are running out of reagents for R9 kits that Nanopore no longer makes.

If you need any extra R10 data, we have Gbps of thaliana data with R10. Thanks!

PengNi commented 10 months ago

@drfultz Thank you very much for using our tool! We are now developing deepsignal3 and training new models for R10 data of plants in POD5 format. As deepsignal3 is still under activate development, there could be interface changes and changes to default parameters. We hope that we can release a stable version soon!

Thank you also very much for willing to share data to us! I'll report back to you once we make progress.

Best regards, Peng

AlineMuyle commented 10 months ago

Hi @PengNi thank you for the development of deepsignal3, I was wondering if models trained on R9 data will still work in deepsignal3? I would like to use it on some old R9 plant data I have to infer methylation in all contexts (CG, CHG and CHH). I look forward to being able to use deepsignal3 on plant data in all context, please let us know when this becomes possible. Best wishes

PengNi commented 9 months ago

Hi @AlineMuyle , sorry for the delay. We are now training an R10 model using deepsignal3. The R9 model with pod5 as input may come later than the R10 model.

Best, Peng

AlineMuyle commented 9 months ago

@PengNi Great! Thank you for the update.

PengNi commented 6 months ago

@drfultz @jcolicchio-soundag @AlineMuyle , Hi all, very sorry for the delay! Our tool deepsignal3 supports R10.4.1 POD5 format now. Please check the deepsignal3 repo for more details if you are still interested.

Best, Peng