Open duoduoduo-d opened 11 months ago
Hi ! I would like to know also. I think the README should be updated, since it's confusing which model should be used with which scripts. So for the first (more recent) solution where you just run dREG/run_dREG.bsh one should use the asvm.gdm.6.6M.20170828.rdata model as far as I understand.
For the second (older dREG-HD) solution, one should first use the run_predict.bsh script. The README says one should use the dREG_model/asvm.RData model when possible. However, there's no model with this name, but there is one at dREG-Model/asvm.mammal.RData. Should we use this one or the asvm.gdm.6.6M.20170828.rdata ?
Also, is the writeBed.bsh script still needed ? I only get bed files and bw files as outputs.
Hi, apologies for the delayed response. As far as I'm aware, asvm.gdm.6.6M.20170828.rdata
is the most recent model and can be downloaded through the ftp link in the main README (I'll also make a Zenodo archive when I have the chance). This should be run using the run_dREG.bsh
script, per section 1 "Peak Calling" in the main README, and is what is implemented on the dREG gateway. This script/model is designed to be GPU-accelerated and will run very slowly on CPU. If you don't have access to an Nvidia GPU, or have trouble with CUDA dependencies, please submit jobs to the gateway.
Again, sorry for the very late reply. I haven't worked directly on dREG, but I've had a few conversations with Charles and Zhong and will update the primary documentation when I have some free time.
I have many questions. Is run_predict.bsh a more advanced version of run_dREG.bsh? I can't find the dREG_model/asvm.RData (asvm.mammal.RData seems to be a sample dataset). Does this mean that run_predict.bsh also needs to use asvm.gdm.6.6M.20170828.rdata?