RostSpace

RostSpace visualizes embeddings in 3D and 2D space and colors them by groups, provided throug a CSV file. Dimensionality reduction can be performed with UMAP, PCA or t-SNE. Additional features for biologists are implemented, to display molecule structures by providing PDB files.

Installation

RostSpace can be installed with pip:

# install it with pip
pip install rostspace

Getting started

After installation, RostSpace can be run using the rostspace command.

The required arguments are:

-o      Path to the output directory where generated files are stored
--hdf   Path to HDF5-file containing the per protein embeddings as key-value pair
--csv   Path to the .csv-file containing the metadata

Optional arguments are:

--pdb   Path to the directory that holds the .pdb files
--fasta Path to the .fasta-file
-v      Verbose -- prints internal process to the terminal

Download example data

Example data can be downloaded from here.

Run RostSpace

RostSpace can be run by providing arguments on the command line, by giving it a YAML file, or a combination of both.

1. Giving arguments on the command line:

   rostspace -o data/KLK --hdf data/KLK/KLK_esm2.h5 --csv data/KLK/KLK.csv

2. Creating a YAML file:

   Pla2g2.yaml

   o: data/Pla2g2
   hdf: data/Pla2g2/emb_esm2.h5
   csv: data/Pla2g2/Pla2g2.csv

   rostspace -conf conf/Pla2g2.yaml

3. Using a YAML file and giving extra arguments:

   rostspace -conf conf/Pla2g2.yaml --pdb data/Pla2g2/colabfold/pdb

For more information to the arguments run

rostspace --help