pmcarlton/umap-sequence-plotting

Visualization and comparison of protein and DNA sequence embeddings

umap-sequence-plotting

Visualization and comparison of protein and DNA sequence embeddings

Instructions should work on Mac OSX of recent versions (v12 and higher)

The steps needed are:

1. Obtain protein embedding files from Uniprot

2. Prepare the python environment

3. Modify (if necessary) and run the provided script

Obtaining the embedding files:

See the explanation here: https://www.uniprot.org/help/embeddings

Download from here: https://www.uniprot.org/help/downloads

The two files needed for this example are:

• https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/embeddings/UP000001940_6239/per-protein.h5
• https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/embeddings/UP000005640_9606/per-protein.h5

Note that these files have the same base filename, so you'll need to rename them (e.g. "elegans-protein.h5")

Download these files one at a time, renaming as necessary, and place them in a new directory.

Preparing the python environment

1. Install "pip" (a package manager for python) from the Terminal if necessary:

python3 -m ensurepip

2. Install the needed modules (Dash, UMAP, Pandas, and iPython for ease of use):

One way to do it is:

pip install dash
pip install pandas
pip install umap-learn
pip install ipython

Another way is to use conda to create a virtual environment and install the same thing (TODO explain more about this)

3. Start iPython and run the "worm_human_plot.py" script:

ipython

(you can then paste in the commands from the script line-by-line)

Alternatively you can just run the script using python3 worm_human_plot.py from the Terminal, non-interactively.