search

chanzuckerberg / shasta

[MOVED] Moved to paoloshasta/shasta. De novo assembly from Oxford Nanopore reads
Other
270 stars 57 forks source link

ERROR: guppy-5.0.7-a is not a built-in Bayesian model #264

Closed madagiurgiu25 closed 3 years ago

madagiurgiu25 commented 3 years ago

Hi,

According to the documentation https://chanzuckerberg.github.io/shasta/ComputationalMethods.html#IterativeAssembly the guppy-5.0.7-a should be available as Bayesian model. I installed shasta-Linux-0.7.0 from source but when I tried running the assembly with this preset I received the following error:

Command: ./shasta-Linux-0.7.0 --input $input --Assembly.consensusCaller Bayesian:guppy-5.0.7-a

Error: guppy-5.0.7-a is not a built-in Bayesian model and could not be open as a configuration file. Valid built-in choices are: guppy-2.3.1-a, guppy-2.3.5-a, guppy-3.0.5-a, guppy-3.4.4-a, guppy-3.6.0-a, r10-guppy-3.4.8-a

Do I need to install maybe a different version of Shasta? Thank you!

paoloczi commented 3 years ago

As you suspect, this is a feature that was recently added, and it is not in release 0.7.0. The documentation on GitHub Pages always refers to the latest code on GitHub, and the 0.7.0 documentation (available in the 0.7.0 tar file) does not mention this new option. Sorry about the confusion and inconvenience.

I hope to have soon a new release that includes this, but in the meantime you can download a current Shasta test build as described here. For convenience, the link to download a test build from about a week ago is here. You must be logged in to GitHub while using this link. After download you have to unzip and chmod to make it executable, after which you should be able to run it and it should recognize the Bayesian model for Guppy 5.0.7 (which incidentally in my experience creates reads of significantly better quality than was available previously). Let me know if you have any problems doing this.

paoloczi commented 3 years ago

Given that you did not comment I assume that you were able to download a test build that includes the new Bayesian model. Therefore I will close this issue, but feel free to reopen it or create a new one if you have additional question or problems.