Roadmap for 0.1.3 release

[jisraeli]

DragoNN aims to democratize deep learning for genomics by providing resources to both teach and learn about deep learning for genomics. In a twitter poll last month, I asked deep learning practitioners what they need to get started in genomics: The most common answer, 29 out of 79 votes, was "tutorials". In an ongoing twitter poll, I ask computational biology faculty what they need to teach deep learning for genomics in their classes: The most common answer so far, 24 out of 59 votes, is "starter teaching material". The demand is clear and, in collaboration with Nvidia, we took a big step last week to address this demand by debuting an online Nvidia deep learning for genomics class using DragoNN.

For the 0.1.3 release I would like to provide the minimal starter teaching resources here to enable faculty to teach this topic in their classes. Below is an initial set of todos based on feedback so far:

• Easier installation. A pypi release may help in this regard.
• Automated build/image for cloud usage. This is available now on the GTC branch and needs to be merged.
• DragoNN cloud instances for GCP/Azure.
• Tutorials with reasonable runtime on a laptop without a GPU. This is available now on the GTC branch and needs to be merged (same one as in the Nvidia online course).
• Tutorials that show deep learning succeeding where simpler models such as PWMs fail.
• Tutorials with more figures that could be followed without the need for accompanying slides.

Additional suggestions are always welcome so please feel free to comment and discuss!

We would also love contributions of specific features and/or tutorials. It would be great to have a collection of exercises and tutorials here others could reuse.

[jisraeli]

@mguo123 @a80: would be great to make your GCP instance for DragoNN accessible to the community and add to the cloud resources page on the DragoNN site.