Personalized Medicine: Redefining Cancer Treatment

Prediction the effect of Genetic Variants to enable Personalized Medicine

Author: Eda AYDIN - Training Instructor: Hoang Quy La

Business Understanding

A lot has been said during the past several years about how precision medicine and, more concretely, how genetic testing is going to disrupt tthe way diseases are treated.

But this is only partially due to the huge amount of manual work still required. Once sequenced, a cancer tumor can have thousands of genetic mutations. But the challenge is distinguishing the mutations that contribute to tumor growth (drivers) from neutral mutations (passengers)

Currently this interpretation of genetic mutations is being done manually. This is a very-time consuming task where a clinical pathologist has to manually review and classify every single genetic mutation based on evidence from text-based clinical literature.

We need to develop a machine learning algorithm that, using this knowledge base as a baseline, automatically classifies genetic variations.

This problem was a competition on Kaggle. This was launched by Memorial Sloan Kettering Cancer Center (MSKCC)

Data Understanding

There are nine different classes a genetic mutation can be classified on.

This is not a trivial task since interpreting clinical evidence is very challenging even for human specialists. Therefore, modeling the clinical evidence (text) will be critical for the success of your approach.

Both, training and test, data sets are provided via two different files. One (training/test_variants) provides the information about the genetic mutations, whereas the other (training/test_text) provides the clinical evidence (text) that our human experts used to classify the genetic mutations. Both are linked via the ID field.

Therefore the genetic mutation (row) with ID=15 in the file training_variants, was classified using the clinical evidence (text) from the row with ID=15 in the file training_text

Finally, to make it more exciting!! Some of the test data is machine-generated to prevent hand labeling. You will submit all the results of your classification algorithm, and we will ignore the machine-generated samples.

File Descriptions

• training_variants - a comma separated file containing the description of the genetic mutations used for training. Fields are ID (the id of the row used to link the mutation to the clinical evidence), Gene (the gene where this genetic mutation is located), Variation (the amino acid change for this mutations), Class (1-9 the class this genetic mutation has been classified on)
• training_text - a double pipe (||) delimited file that contains the clinical evidence (text) used to classify genetic mutations. Fields are ID (the id of the row used to link the clinical evidence to the genetic mutation), Text (the clinical evidence used to classify the genetic mutation)
• test_variants - a comma separated file containing the description of the genetic mutations used for training. Fields are ID (the id of the row used to link the mutation to the clinical evidence), Gene (the gene where this genetic mutation is located), Variation (the amino acid change for this mutations)
• test_text - a double pipe (||) delimited file that contains the clinical evidence (text) used to classify genetic mutations. Fields are ID (the id of the row used to link the clinical evidence to the genetic mutation), Text (the clinical evidence used to classify the genetic mutation)
• submissionSample - a sample submission file in the correct format