Evaluation of deep learning techniques for identification of sarcoma-causing carcinogenic mutations

Area of Science:

• Genomics
• Computational Biology
• Oncology

Background:

• Sarcoma is a significant cancer type affecting bones and soft tissues, predominantly in children.
• Early cancer detection is crucial for improving patient outcomes and survival rates.
• Genetic mutations in driver genes are implicated in sarcoma development.

Purpose of the Study:

• To develop a deep learning framework for the early detection of human sarcoma cancer.
• To leverage Recurrent Neural Network (RNN) algorithms for analyzing gene mutation data.
• To identify potential genetic markers for sarcoma diagnosis.

Main Methods:

• Utilized a dataset comprising 141 mutations across 8 driver genes from over 134 samples.
• Trained Long Short-Term Memory (LSTM), Gated Recurrent Units, and Bi-directional LSTM (Bi-LSTM) algorithms on gene sequences.
• Applied rigorous validation techniques including self-consistency, independent set testing, and 10-fold cross-validation.

Main Results:

• The proposed deep learning model achieved a high accuracy of 99.6%.
• An Area Under the Curve (AUC) value of 1.00 was obtained, indicating excellent model performance.
• Key metrics such as sensitivity, specificity, and Mathew's correlation coefficient were evaluated.

Conclusions:

• The developed RNN-based framework demonstrates high efficacy for early sarcoma cancer detection.
• This approach holds promise for improving diagnostic accuracy and patient stratification.
• Further research can explore larger datasets and diverse gene mutations for enhanced predictive power.