Day 1 // Hands-on 2 // Explore SingleCellExperiment

Load packages

library( "Seurat" )
library( "scater" )
library( "scran" )

library( "tidyverse" )

Load data

Seurat -> SingleCellExperiment

sce <- readRDS( "/home/idies/workspace/practical_genomics/day1/retina.rds" )
sce <- as.SingleCellExperiment( sce )

Check data

1. assay()

Introduce

• How to access the expression data?

help( assay )

m_assay <- assay( , "logcounts" )

• What class of object is this?

class(  )

• How large is it?

dim(  )

• How to subset first 10 rows, first 3 columns?

m_assay[ ,  ]

Subset RHO

x_rho <- m_assay[ , ]

( x_rho )

( x_rho )

hist(  )

summary(  )

Subset RHO, VIM

goi <- c( "RHO", "VIM" )
x_rhovim <- m_assay[ , ]

( x_rhovim )

( x_rhovim )

( x_rhovim )

ggplot RHO, VIM

• t(), data.frame(), rownames_to_column(), pivot_longer()

df_rhovim <- ( x_rhovim )
df_rhovim <- ( df_rhovim )
df_rhovim <- ( df_rhovim )
df_rhovim <- ( df_rhovim, cols=c( "RHO", "VIM" ) )
df_rhovim

ggplot(  ) +
  geom_histogram( aes(  ) ) +
  facet_wrap( vars() )

2. colData()

Introduce

• How to access the cell metadata?

( colData )

meta <- colData(  )

• What class of object is this?

( meta )

• What is DFrame?

meta
data.frame(  )

• How large is it?

( meta )

• What categories exist?

colnames(  )

Subset rows

• colData(), data.frame(), rownames_to_column()

df_meta <- ( sce ) %>%
  () %>%
  ()
df_meta

( df_meta )

df_meta[ ,  ]

Subset libraryID

table( df_meta$ )

df_fovea2 <- df_meta %>%
  filter( libraryID=="" )
df_fovea2

( df_fovea2 )

table( df_fovea2$ )

ggplot gene by cell_type

• full_join(), filter()

df_rhovim_fovea2 <- df_rhovim %>%
  ( df_meta, by="rowname" ) %>%
  ( libraryID=="fovea_donor_2" )
df_rhovim_fovea2

• geom_violin(), facet_wrap(), theme()

ggplot( df_rhovim_fovea2 ) +
  ( aes( cell_type, value ) ) +
  ( vars(name) ) +
  ( axis.text.x=element_text( angle=90 ) )

3. reducedDim()

Introduce

• How to access the cluster visualization?

( reducedDim )

• data.frame(), rownames_to_column()

df_umap <- reducedDim( sce, "" ) %>%
  () %>%
  ()
df_umap

ggplot cell_type

df_umap_meta <- full_join( , , by="rowname" )
df_umap_meta

ggplot(  ) +
  geom_point( aes( , , color= ) )

ggplot RHO

df_umap_rho <- full_join( , , by="rowname" ) %>%
  filter( name=="" )
df_umap_rho

ggplot(  ) +
  geom_point( aes( , , color= ) )

4. findMarkers()

Introduce

• How to access differentially expressed genes?

( findMarkers )

• Preview of findMarkers() syntax

colLabels( sce ) <- colData( sce )$
markers <- findMarkers(  )

• What class of object is this?

( markers )

• How large is it?

( markers )

• Build up to data.frame( markers[[1]] )

markers

hist summary.logFC

markers[1]

df_markers_amacrine <- markers[[1]] %>%
  ()
df_markers_amacrine

hist( df_markers_amacrine$ )

plotUMAP markers

df_markers_amacrine %>%
  filter( summary.logFC >  )

help(  )

plotUMAP( sce, colour_by="MEG3", ="cell_type" )

Exercise 2

1. Recreate plotUMAP() using ggplot2
   • Plot another gene, cell metadata
2. Recreate plotExpression() using ggplot2
   • Plot another gene by cell metadata
3. Adjust ggplot2 appearance
   • Add titles, customize x/y labels, theme
4. Explore findMarkers() results
   • Use plotUMAP(), plotExpression()

Document software

sessionInfo()