The First Step in Oncogenicity Determination - Predict the effect of Tumor Genetic Variants on Protein Function.

Which of the thousands of mutations in a cancer tumor contribute to the tumor growth. This project and Kaggle competition will uses machine learning techniques to evaluate vast amounts of an expert-annotated knowledge base of information about each mutation to determine which of the thousands of mutations cause tumor growth.(1)

Click above Capstone.pdf to see report. Click above Capstone_mlnd_Unabriged_all_code_very_long.ipynd to see report including all code.

Additional gene function and pathway information was downloaded from "Database for Annotation Visualization and Integrated Discovery"(DAVID) to a david_full.csv file and is included in the submissions.

Also additional gene summary data was downloaded from "National Center for Biotechnology Information"(NCBI). A gene symbol / gene name /gene id conversion function on the David website was used. The unique gene symbol/ name / id is include included gene_id.csv

The env.yml included above, was the environment used.

To view report click Capstone.pdf above.

To run code without the additional biological data:

• The data can be downloaded directly form Kaggle at https://www.kaggle.com/c/msk-redefining-cancer-treatment/data
• Clone notebook above.
• Data needs to be in the same directory as the jupyter notebook.

To run code with additional biological data:

• The data can be downloaded directly form Kaggle at https://www.kaggle.com/c/msk-redefining-cancer-treatment/data

• Clone notebook above.

• Data needs to be in the same directory as the jupyter notebook.

• An additional library was used http://biopython.org/

biopython was used and from Bio 'Entrez' was imported.

This was difficult to set up.

The jupyter notebook should be able to run now top to bottom as is with running the additional biological code.

The Capstone report is a pdf. The code to reproduce all results and images are in a jupyter notebook. Since the additional libraries were difficult to set up, all related code for the additional biological data is commented out so the jupyter notebook can be run easily from top to bottom. This will produce all other results. The pdf report will include information about the additional biological data.

Gene id was needed to search the ncbi website. The given data used gene name. Luckily the biological website david.com has a conversion function. So I created a uniq_genes.csv up loaded it to the david site and it covered the gene names to gene id and it was downloaded to my computer as gene_ids.csv. This file is also included in my submission. The Entrez web site were I did a search required my email address. This is in my code. If you choose to run the additional biology code you may need to change this to your email. I think it should work with my email.

The david_full.csv, uniq_genes.csv and the input data file(from kaggle, the address is above) all need to be in the same directory as the jupyter notebook.