search

bowang-lab / BIONIC

Biological Network Integration using Convolutions
MIT License
59 stars 4 forks source link

Large network data integration #44

Closed zhanlingm closed 2 years ago

zhanlingm commented 2 years ago

Hello, Doctor! I have learned a lot from your article recently, which just solved the problem I encountered. However, there is still a problem. When I used the bionic method to deal with a large network, the processing effort was extremely long. Of course I use the CPU processing, so want to consult, for the large network should be how to deal with, how the computer configuration?

Thank you so much!

duncster94 commented 2 years ago

Hi, thanks for opening an issue. What version of BIONIC are you using? I recently released v0.2.5 which significantly improves the network preprocessing.

zhanlingm commented 2 years ago

Hi, thank you for taking time out of your busy schedule to reply to my message. I am using bionic v0.2.5. When I used bionic to run pathway data (11167 nodes and 100,000 relationship edges), the total time was 108 days. Is it because of my CPU configuration?

Hi, thanks for opening an issue. What version of BIONIC are you using? I recently released v0.2.5 which significantly improves the network preprocessing.

duncster94 commented 2 years ago

Hi, it sounds like the training is what's taking a long time. Yes, the CPU version is only suitable for small networks. I recommend you install a CUDA version and run BIONIC on a GPU.

zhanlingm commented 2 years ago

Hi, it sounds like the training is what's taking a long time. Yes, the CPU version is only suitable for small networks. I recommend you install a CUDA version and run BIONIC on a GPU.

hi, thank you for your patience and guidance. I got a lot of help. Thank you again!

duncster94 commented 2 years ago

No problem! I'll close this issue but if you run into problems feel free to reopen it or open a new one.