Wei Zhang, wz2363 | Lingyi Cai, lc3352

• Introduction video link

• Application link

• Model source code

• Web source code

Python: 3.6.12

Pytorch: 1.7.0

R: 4.0.5

From studies from the Centers for Disease Control and Prevention (CDC), Each year in the United States, more than 1.7 million people are diagnosed with cancer, and almost 600,000 die from it, making it the nation’s second leading cause of death. The cost of cancer care continues to rise and is expected to reach almost $174 billion by 2020, it's also very shocking that 1 in 3 people will have cancer in their lifetime. As we know cancer is complicated to cure, more than one thousand genes may involve in cancer development, and there are thousands of subtypes of cancer. So, we will be trying to make a real and tangible difference in the fight against cancer. In addition, we propose to implement a web server for this framework, so that the analysis can be performed without writing any code. This user-friendly tool can be convenient for biologists who are not familiar with programming languages. From studies from the Centers for Disease Control and Prevention (CDC), Each year in the United States, more than 1.7 million people are diagnosed with cancer, and almost 600,000 die from it, making it the nation’s second leading cause of death. The cost of cancer care continues to rise and is expected to reach almost $174 billion by 2020, it's also very shocking that 1 in 3 people will have cancer in their lifetime. As we know cancer is complicated to cure, more than one thousand genes may involve in cancer development, and there are thousands of subtypes of cancer. Meanwhile, the deep learning technology has been revolutionizing over the past decade in many fields such as computer vision, natural language processing and many others. However, there are two challenges remaining. First, the black-box property of deep neural networks has made models hard to explain and analyze. Tackling this challenge becomes even harder as the architectures of neural networks are deriving intricately nowadays. Second, although deep neural networks have achieves extraordinary performance in tasks such as classification, such calibrated designs might only learn the association instead of the causation. Those two remaining challenges hinder the application of deep learning technologies from being used in a wider range such as financial industry and healthcare system. In addition, we propose to implement a web server for this framework, so that the analysis can be performed without writing any code. This user-friendly tool can be convenient for biologists and doctors who are not familiar with programming languages.