ISTDECO: In Situ Transcriptomics Decoding by Deconvolution

In Situ Transcriptomics (IST) is a set of image-based transcriptomics approaches that enables localisation and quantification of gene expression directly in tissue samples. IST techniques usually produce large multiplexed image series in which fluorescent spots are either present or absent across imaging rounds and colour channels. A spot's presence and absence in the different rounds and colour channels forms a barcode. IST experiments are usually designed such that different types of barcodes correspond to different types of mRNA. Therefore, the expression of a gene can be quantified by localising the fluorescent spots and decode the barcode that they form. Existing IST algorithms usually carry out these assessments in two separate steps: spot localisation and barcode decoding. Although these algorithms are efficient, they either assume that spots are sufficiently separated for an adequate localisation, or that the barcoded intensities are error-free for a correct decoding. Both these assumptions are likely to be violated when the spot density is high or in regions of low signal-to-noise ratio. We argue that an improved gene expression decoding can be obtained by combining the localisation and barcode decoding into a single algorithm. This allows for an efficient and intuitive gene expression decoding that is less sensitive to noise as well as optical crowding. We offer In Situ Transcriptomics Decoding by Deconvolution (ISTDECO), a principled decoding approach combining spectral and spatial deconvolution into a single algorithm. We evaluate ISTDECO on synthetic data at varying noise ratios and signal densities, as well as on two real world datasets, and compare with existing state-of-the-art. ISTDECO achieves state-of-the-art performance despite high spot densities and low signal-to-noise ratios. It is easily implemented in just a few lines of code and runs very efficiently using a GPU.