Patient1

• 

  Joining 120 bio researchers and data scientists to collaborate on John's genomics research case

• 

  We created a Colab notebook with Tensorflow2 to give researchers a head start analyzing John's data

• 

  It can be used to detect genetic variants that can help in finding a diagnosis and treatment.

• 

  We upgraded the Clairvoyante variant caller to work with Tensorflow 2

How can Tensorflow 2 help patients with undiagnosed chronic diseases?

John is a 33-year-old Caucasian male suffering from one or more mysterious genetic diseases. Since childhood, he has had a significant challenge in gaining weight despite adequate caloric intake. As a child, he reported nausea, stomach aches and an overall aversion to food. He developed chronic daily vomiting and would vomit as many as 5 times per day. At his current age, John is 5’10” tall and weighs 109lbs. He is easily fatigued due to his limited muscle mass and low weight.

At Undiagnosed-1 in June we will join 120 others from SVAI (Silicon Valley Artificial Intelligence) over a weekend to look for answers and create medical insights into John's unresolved condition. Armed with $150K in Google cloud credits, we will apply data science on his medical data including Genome, Exome, and Proteomics to generate ideas for diagnosis, explore treatment options and identify how he can better manage his symptoms.

In preparation for the event, we have created a Tensorflow 2 based Colab notebook with tools to analyze John's health data. The open source code is made available for participants to give them a head start on the data analysis in advance of the in person event.

For the notebook, we upgraded the Clairvoyante variant caller to work with Tensorflow 2. It detects differences in an individual genome relative to a reference sequence, and is implemented as a multi-task five-layer convolutional neural network model for predicting variant type, zygosity, alternative allele and Indel length.

Built With

• clairvoyante
• colab
• python
• tensorflow