Installation instruction

as of Nov 2020; will be cleaned up in the future

Docker

• Install Docker
  • Enable running docker without sudo (see this; if running on lab server and don't have sudo privilege, ask Loren)
  • test with: docker run --rm hello-world

Database access

• Ask Loren for mysql access to lmf-db.cin.ucsf.edu
• If you're looking to practice (recommended if starting out), you can use your own database (see this)

Create a conda environment

• conda env create --name nwb_datajoint python==3.8

From Anaconda / PyPI:

• Install nodejs
  • conda install nodejs -c conda-forge --repodata-fn=repodata.json
• Install jupyterlab
  • conda install -c conda-forge jupyterlab
• Install hither
  • pip install hither
• install spikeinterface
  • pip install spikeinterface
• Install ml_ms4alg
  • pip install ml_ms4alg
• Install datajoint
  • pip install datajoint

From GitHub:

• Install LorenFrankLab/nwb_datajoint
  • python setup.py develop
• Install NovelaNeuro/ndx-franklab-novela
• Install Labbox-ephys
  • install the launcher as well as the package
  • setup.py is in the python subfolder

Configure Datajoint

• Change your password for accessing Datajoint database and set up external stores. Should need to run these only once.

Notes

• Add environment variables (e.g. in ~/.bashrc); note that these assume that you're interacting with DJ database from one of the lab's servers (e.g. virgas)
  • export LABBOX_EPHYS_DATA_DIR="/stelmo/nwb/"
  • export NWB_DATAJOINT_BASE_DIR="/stelmo/nwb/"
  • export KACHERY_STORAGE_DIR="/stelmo/nwb/kachery-storage"
  • export KACHERY_P2P_CONFIG_DIR="/your-home-directory/.kachery-p2p"
  • export DJ_SUPPORT_FILEPATH_MANAGEMENT=true
  • export KACHERY_P2P_API_PORT="some-port-number" (this is optional)
• For curation with the web GUI, you must be running kachery-p2p daemon and labbox-ephys in the background
  • kachery-p2p: kachery-p2p-start-daemon --label franklab --config https://gist.githubusercontent.com/khl02007/b3a092ba3e590946480fb1267964a053/raw/f05eda4789e61980ce630b23ed38a7593f58a7d9/franklab_kachery-p2p_config.yaml
  • labbox: see instruction here
  • if running jupyter widget, see instruction here

Troubleshooting common problems

• If cannot connect to dj, then downgrade pymysql to 0.9.2
• If cannot write nwb, then downgrade h5py to 2.10.0
• If stall during spikesorting, then install ml_ms4alg from our fork and upgrade numpy to 1.19.4
• If cannot get waveforms, downgrade following packages to these versions:
  spikecomparison 0.3.0
  spikeextractors 0.9.1
  spikefeatures 0.1.1
  spikeinterface 0.10.0
  spikemetrics 0.2.2
  spikesorters 0.4.2
  spiketoolkit 0.7.0
  spikewidgets 0.5.0