Developing software for compressed imaging transcriptomics

Abstract

Modern-day biological experimentation often necessitates a scale of data that is exponential with respect to the number of genes that are being measured, and this in turn leads to high latency and monetary cost during hypothesis testing. In addition to such practical constraints, some biological experiments are just physically infeasible due to fundamental limitations on the throughput of current technologies. However, because nearly all biological data are highly structured and can be described in terms of relatively few components, it is not necessary to measure each data point individually. Instead, using the framework of compressed sensing, it is possible to take advantage of this structure to gather the requisite data for an experiment while collecting only a fraction of the original number of measurements. In previous work, we have applied compressed sensing for the particular purpose of generating spatial gene expression profiles using fluorescence microscopy (i.e. imaging transcriptomics). In order to make this technique more accessible and user-friendly, we built CISIpy, an open-source software system that implements the pipeline's computational aspects. This system is designed to enable efficient compressed sensing workflows that is highly portable across platforms and especially amenable to cloud computation. The end result is a well-tested, open-source software package replete with functionality, documentation and examples.