-
Virtual Systematic Gene Knockout in scRNAseq Data
Introduction A few years ago I came across this paper by Michael W. Dorrity and Lauren M. Saunders et. al. who used dimensionality reduction (DR) techniques to infer protein complexes and pathways from a dataset of 1,484 single gene deletions in the yeast genome. They used a DR algorithm called...
-
Shapley Values for Cytometry Data Interpretation
Introduction Dimensionality reduction techniques help us condense high-dimensional data, however these can sometimes be difficult to interpret, especially when you try to consider how much each feature actually contributes to the result. In the field of explainable AI, Shapley values are frequently used to interpret the feature importance of a...
-
Animating tSNE iterations
Environment Setup library(dslabs) library(rsvd) library(dplyr) library(tidyverse) library(gganimate) source('~/development/FIt-SNE-master/fast_tsne.R', chdir=T) set.seed(1234) Data Import and Preprocessing mnist <- read_mnist() Stepwise tSNE Iteration Function tsne_iterate <- function(data, iteration_sequence) { start <- Sys.time() tsne_df_tmp <- data.frame() run <- 0 for (i in iteration_sequence) { run <- run + 1 message(paste('Run', run, 'of', length(iteration_sequence))) tsne_tmp...
-
RNA-seq Data Analysis in R
Environment Setup library(Seurat) library(magrittr) library(dplyr) Data Import and Preprocessing Read data generated by cellranger count/aggregate. Create seurat object, filtering out all genes expressed in less than 3 cells and all cells with less than 200 expressed genes raw_data <- Read10X(data.dir = '.../filtered_gene_bc_matrices/hg19/') cds <- CreateSeuratObject(raw_data, min.cells = 3, min.features =...
-
Flow-Cytometry Data Analysis in R
Flow Cytometry Data Flow cytometry is a powerful tool to analyse cells based on their size, granularity and expression of various intracellular and membrane bound proteins. Thereby, the datasets it produces are multidimensional, with thousands of cells and protein expression levels for each of these. As flow cytometry relies on...
-
Flow-Cytometry Shinyapp
Flow cytometry data is multi-dimensional. Each cell is analysed for its expression pattern of various proteins. Thereby, the data is inherently complex and difficult to clearly understand. Over the past few years dimensionality algorithms have been developed and have gained popularity in the analysis of biological data. This shiny app...