Human beings need help interpreting omic-scale biological data. Safeweb provides 20,000-foot and ground-level views.
Biology is increasingly computational. As scientists work with genome-scale data—say, RNA-seq or CRISPR experiments—they need tools to help them understand what their experiment meant.
For non-programming scientists, contemporary web technologies present the opportunity for accessible, easier-to-use apps. These extensible apps can help biologists understand and generate hypotheses about their data. Safeweb is one such app, for visualization and exploratory analysis of data on biological networks.
I designed and developed Safeweb, primarily in React and Python. It features several functional enrichment algorithms and visualization tools that help scientists get a birds eye view of what their genomics experiments mean biologically, and explore individual hits of interest.
Safeweb was first developed in late 2019 as an interactive, no-programming-required implementation of Anastasia Baryshnikova’s SAFE (spatial analysis of functional enrichment) algorithm, for use with an established yeast genetic interaction network based on double-knockout mutant strain fitness data. We wanted to extend its utility by generating human gene-gene similarity networks.
Gene-gene similarity networks group genes by shared function in the underlying data used to generate the network; even if we don't know exactly why or how, the network allows us to see that genes were correlated in their function. There may be regions of the network that lack coherent functional annotation, but even if a gene's function is uncharacterized, knowing about its neighbors may help with hypothesis generation. Aspirationally, networks allow us to capture the whole of gene function without hiding the parts. In an interactive tool like Safeweb, network visualizations provide both a 20,000-foot (or gene) functional view and the capacity to interrogate individual genes and their connections.