search

epi2me-labs / wf-transcriptomes

Other
76 stars 32 forks source link

Denovo removed? #32

Closed NStrowbridge closed 1 year ago

NStrowbridge commented 1 year ago

Ask away!

Just wondering why denovo pipeline for wf-transcriptomes was removed from the github? Was there issues with denovo option?

Cheers,

Nic

cjw85 commented 1 year ago

Over the lifetime of the workflow, the de-novo method has led to more than its fair share of bug reports and support issues. Whilst some of these were trivially our fault (dumb bugs in the code), many were more algorithmic in nature. That is to say the methods employed were not robust to many people's real world data.

The de-novo method was always documented as being experimental. Because of the various issues encountered we took the hard decision to remove the de-novo option from the workflow. Whilst this represents a loss of functionality, we believe that it is for the better good as the workflow did not work reliably and its results could not always be trusted.

@ksahlin and his group do a lot of good work in this area, and I might direct you to https://github.com/aljpetri/isONform if the de-novo construction of transcriptomes is something of interest to you.

NStrowbridge commented 1 year ago

Chris,

Thanks for the swift reply. I thought as much, with regards to the experimental nature and issues. Thanks for pointing me in the direction of isONform. I will read into that and see if it fits my needs.

Cheers,

Nic

ksahlin commented 1 year ago

Hi @NStrowbridge,

Our reference-free pipeline consists of (pychopper (by ONT), isONclust, isONcorrect, isONform ). Here is a full workflow script. https://github.com/aljpetri/isONform/blob/master/full_pipeline.sh Feel free to reach out if you have any questions.

NStrowbridge commented 1 year ago

Hi @ksahlin,

Thanks for the links, I will be in touch if I have any issues.

Cheers,

Nic