sc-SynO: a LoRAS integration to annotate rare-cell populations in scRNAseq data

Projects that do not fall under current programmes.

The project addresses the generation and establishment of programmed pacemaker cells for an in vitro drug testing possibility to perform predictive tests. This may lead to an improved treatment of cardiac arrhythmias or an accurate identification of potential drug molecules at a very early stage of development. Important benefits will arise in verifying the safety of a wide variety of medicines while reducing animal testing. For more information you may visit our project website at ...

Visualization of the workflow demonstrating a step-by-step explanation for a sc-SynO analysis. a) Several or one snRNA-Seq or scRNA-Seq fastq datasets can be used as an input. Here, we identify our cell population of interest and provide raw or normalized read counts of this specific population to sc-SynO for training. b) Further information for cluster annotation and processed count data are serving as input for the core algorithm. c) Based on the data input, we utilize the LoRAS synthetic ...

Validation of the sc-SynO model for the first use case of cardiac glial cell annotation. UMAP representation of the manually clustered Bl6 dataset of Wolfien et al. (2020) Precicted cells of sc-SynO are highlighted in blue, cells not chosen are grey. UMAP representation of the manually clustered dataset of Vidal (2019). Precicted cells of sc-SynO are highlighted in blue, cells not chosen are grey. Average expression of the respective top five cardiac glial cell marker genes for both validation ...

Validation of the sc-SynO model for the second use case of proliferative cardiomyocytes annotation. a) UMAP representation of the manually clustered single-nuclei dataset of Linscheid et al. (2019) Precicted cells of sc-SynO are highlighted in blue (based on top 20 selected features in the training model), red (based on top 100 selected features in the training model) cells not chosen are grey. b) UMAP representation of the manually clustered dataset of Vidal et al. (2020). PPrecicted cells of ...

Here, we describe the index file generation of the mm10 genome, the genome alignment with kallisto, and quantification with bustools to obtain the used spliced / unspliced transcript input.

Here is the detailed R script to generate the input needed by scSynO for synthetic cell generation and classification model training.

The code that can be embedded into any other Seurat data processing workflow is:

cell_expression_target_cluster <- as.matrix(GetAssayData(seuratobject, slot = "data")[, WhichCells(seuratobject, ident = "target_cluster_number")]) cell_expression_all_other_clusters <- as.matrix(GetAssayData(seuratobject, slot = "data")[, WhichCells(seuratobject, ident = ...

Single nuclei transcriptomics data as .csv files from the Allen Brain atlas data set of mus musculus (https://celltypes.brain-map.org/) have been utilized as an input for scSynO. The underlying analysis is part of the manuscript entitled "Automated annotation of rare-cell types from single-cell RNA-sequencing data through synthetic oversampling". Data anaylsis and visalizations were mainly generated with the Seurat R package (https://satijalab.org/seurat/archive/v3.2/spatial_vignette.html)