VarSome Genome Browser by Saphetor

On April 14th, 2003, the Human Genome Project announced the successful sequencing of approximately 20,000 human genes. Since then, genomics has profoundly impacted public health. Despite the immense progress in the field, extracting insights from the vast expanse of genomic data remains a complex task, requiring meticulous investigation of multiple interdependencies alongside annotations from various sources.

To catalyze the process, we have developed the VarSome Genome Browser, which offers unparalleled efficiency in visualizing genomic data and annotations through a user-friendly web UI. It is highly modular, with four different instances of the core library deployed so far. Users can parameterize colour palettes and themes mainly to accommodate visual deficiencies. Navigation is intuitive, allowing seamless zoom to any level of detail using touch gestures or scrolling.

We built the browser with WebGL, but unlike other 3D graphics visualization tools, the rendered elements have to adjust to the zoom level in specific dimensions (e.g., content width) while remaining constant in others (e.g., padding, height), ensuring that all elements are aligned to their genomic start & end positions while being visible regardless of their size and zoom level. We implemented custom GLSL shaders and instanced buffer geometries to accomplish this vital for the product spatial distortion. A custom coordinate grid system was additionally implemented to rectify 32-bit float precision errors when navigating the full scope of the human genome (length of 3 billion DNA base pairs). A bespoke ray-casting module was engineered to power a highly interactive UI by instantly detecting, in O(1) complexity, the hovered or clicked elements among millions of irregularly shaped and spatially distorted counterparts.

User and system events, API requests, and algorithmic outputs are piped through precisely timed, data streams that orchestrate asynchronous processing workflows and hydrate component states. In conjunction with GPU hardware acceleration, asynchronous execution allows the browser to remain lightweight, maintaining a maximum frame rate even while rendering hundreds of megabytes of data.

Consequently, clinicians can for the first time, in most cases, view a patient’s complete set of genetic variations or gene-wide annotations at once, swiftly discovering patterns and instantly detecting regions of interest through the relevant filters.

The VarSome Genome Browser aids thousands of clinicians and researchers in enhancing diagnostic accuracy, revolutionizing life saving treatment procedures and propelling genomics research toward a future where conditions like rare diseases and cancer are treated before the first symptom emerges.