Notebooks

The notebooks in this folder can be executed locally on your machine or on Google Colab (a tool that allows you to run code online). If you run the Notebooks on your own machine, you might ignore the “Colab Initialization” code, but you will have to download the files required by the notebook. If you do run the Notebooks on Colab, you have to execute commands to install the pipeline and download neccessary files, which are the blocks of code followin the “Colab Initialization” header.

Preface

From experience within our lab and with collaborators we have created a set of Notebooks that try to address different aspects of what is generally needed. The Notebooks presented here are to be viewed as “building blocks” for your exploratory projects! We’ve tried to keep the notebooks short and to the point. Often, you will need to grab a thing from here and a thing from there.

From the manuscript

Purpose	Colab	Notebook
Basic Protocol 2 and alternates: use deeploc embeddings produced by the pipeline to plot sequence spaces. This is virtually the same as this pipeline example, but here we can tune the UMAP parameters until we obtain a nice graphic to put in a presentation :) .	Click here	Click here
Basic Protocol 3: train a simple machine learning classifier to predict subcellular localizations training on DeepLoc embeddings.	Click here	Click here

Exploring modules from the bio_embeddings package

Purpose	Colab	Notebook
Use the general purpose embedding objects to embed a sequence passed as string	Click here	Click here
Use Bio to load a FASTA file and the general purpose embedding objects to embed sequences from the file	Click here	Click here
Embed a sequence and extract annotations using supervised models from bio_embeddings.extract. You can achieve the same results using the pipeline like in this example.	Click here	Click here
Embed a sequence and transfer GO annotations using unsupervised techniques found in bio_embeddings.extract (vistually the same as goPredSim). You can achieve the same results using the pipeline like in this example.	Click here	Click here

Proper use cases from collaborations

Purpose	Colab	Notebook
Embed a few sequences and try out different ideas to see if the embeddings are able to cluster different sequences	Click here	Click here

Exploring pipeline output files

Purpose	Colab	Notebook
Open an embedding file, the principal output of a pipeline run	Click here	Click here
Use embeddings produced by the pipeline to plot sequence spaces. This is virtually similar to using the project and visualize steps of the pipeline, but in this case, you can change paramteres on the fly in the Notebook. A similar pipeline example here.	Click here	Click here
Use deeploc embeddings produced by the pipeline to plot sequence spaces. This is virtually the same as this pipeline example, but here we can tune the UMAP parameters until we obtain a nice graphic to put in a presentation :) .	Click here	Click here
Analyze embedding sets by studying their similarity and transferring annotations	Click here	Click here
Naïvely plot embeddings to distinguish patterns in your embedded sequences	Click here	Click here

Advanced use cases

Purpose	Colab	Notebook
Train a simple machine learning classifier to predict subcellular localizations training on DeepLoc embeddings.	Click here	Click here

Utilities

Purpose	Colab	Notebook
Re-index an embeddings_file using your original identifiers. Useful when creating reference embedding sets for the unsupervised extract stage	Click here	Click here
Remove identifiers from an annotation file, useful when the pipeline suggest you to do so! :)	Click here	Click here