Optimization of UMI counting strategies for Spatial Transcriptomics
Year: 2018
Type: dissertation
Abstract: Spatial Transcriptomics (ST) is a microarray-based RNA sequencing technology that allows for genome-wide transcriptome profiling of tissue sections with spatialresolution, which was published in Science by Ståhl and Salmén et al in 2016. Polyadenylated transcripts are captured on a microarray surface through hybridizationwith a barcoded DNA oligo that carries the information necessary to infer spatialposition and transcript uniqueness from the output sequencing reads. The ST protocolutilizes Unique Molecular Identifiers (UMIs) to remove PCR duplicates and obtain reliable estimates of gene counts. However, errors gradually accumulate in the UMIpool as a consequence of enzymatic conversion steps and these errors ought to be addressed computationally in data processing steps to produce reliable gene countestimates.In this thesis, we used ERCC reference RNAs and a set of custom-made DNA oligosas spike-ins in the ST protocol to explore sources of technical variation under controlled experimental conditions. An exploratory analysis of the spike-in data gave new insights into the chemistry which could guide future improvements of the protocol. We also developed a new strategy to bin read alignments before gene quantification and showed that this strategy produces a more reliable output. Finally, we developed an in silicosimulation of the ST library preparation to provide a framework that can be used toevaluate the performance of various computational processing strategies. The simulation was then used to benchmark a set of duplicate removal algorithms used toquantify gene expression.This master thesis project was carried out at SciLifeLab at the division of GeneTechnology under supervision of José Fernandez Navarro.
Author: Ludvig Larsson
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W3026583302