Dual spatially resolved transcriptomics for SARS-CoV-2 host-pathogen colocalization studies in humans
Hailey Sounart, Enikő Lazar, Yuvarani Masarapu, Jian Wu, Tibor Várkonyi, Tibor Glasz, András Kiss, Erik Borgström, Zsuzsanna Varga, Olaf Bergmann, Stefania Giacomello
The list of SARS-CoV-2 genes targeted by the probes in this study and used in the data analysis can be found in this csv file. The sample IDs for corresponding count matrices can be accessed here.
The counts can be accessed here. Corresponding sequences fastq files can be accessed upon request from here. High-resolution tissue images can be downloaded from here.
Scripts used for generating count matrices can be accessed under spaceranger-scripts folder.
All R scripts used to run the human genes clustering analysis can be found in sequential order ("<#>_filename.Rmd") under folder R_scripts. This also includes the colocalization analysis of the SARS-CoV-2+ and SARS-CoV2- spots that was performed and mentioned in the manuscript. The metadata used in the 1_add_metadata.Rmd is within this sheet.
Deconvolution was performed using Stereoscope. The input single cell data for the deconvolution was the SCP1052 dataset taken from the study Delorey, T.M., Ziegler, C.G.K., Heimberg, G. et al. COVID-19 tissue atlases reveal SARS-CoV-2 pathology and cellular targets. Nature 595, 107–113 (2021). https://doi.org/10.1038/s41586-021-03570-8.
This dataset was subsampled to include donor ids: D1, D4, D5, D6, D8, D12, D18, D14, D16. These donor ids were selected based on the timeframe of symptom onset to death withon 13-20 days which is close to the duration of 13-17 days from diagnosis to death in our study.
Two Stereoscope runs were performed, one setting was tested for each run. These are stated below.
Setting 1: 5000 sc epochs and 10000 st epochs
Setting 2: 15000 sc epochs and 25000 st epochs
The single cell data subsampling and preparation of the ST dataset for deconvolution was performed using scripts prep_sc_data_deconv-CovidLung.Rmd and prep_st_data_for_deconv-CovidLung.Rmd respectively. The results of the deconvolution were summarised in the script summary_stereoscope_covid-lung.Rmd.
More extensive analysis based on deconvolution to identify cluster-specific and cell-type-specific DE genes between the SARS-CoV-2+ and SARS-CoV2- spots was performed using the script colocalisation_DE_deconvolution-based.Rmd.
Analysis of the SARS-CoV-2+ spots to identify covid genes co-occurance with each other and their distribution can be re-run using the file covid-genes_co-occurance.Rmd
The Fisher's test and the Chi-square tests for the covid genes localization were performed using the script covid_colcoalization_pvalues.Rmd and the file covid_gene_colocalization.csv.
All the required scripts for running elastic registration, RNAscope and ST calls, validation analysis and running chi-square test on the confusion matrix are available in the folder RNAScope validation.
There is a ReadMe.txt file in the same folder that describes what each script does in more details that are required for reproducibility.
The chi-square test on the confusion matrix was performed using the script [confusion_matrix_pvalue.Rmd](RNAScope validation/confusion_matrix_pvalue.Rmd).