Combining Deep Learning and BCRs with Ibex

Getting Started

The idea behind Ibex is to combine BCR CDR3 amino acid information with phenotypic RNA/protein data to direct the use of single-cell sequencing towards antigen-specific discoveries. This is a growing field - specifically TESSA uses amino acid characteristics and autoencoder as a means to get a dimensional reduction. Another option is CoNGA, which produces an embedding using BCR and RNA. Ibex was designed to make a customizable approach to this combined approach using R.

More information is available at the Ibex GitHub Repo.

Installation

devtools::install_github("ncborcherding/Ibex")

The Data Set

To show the multiple options of Ibex, the example data is derived from this manuscript, multimodal single-cell characterization of COVID19-associated multisystem inflammatory syndrome in children.

Formation of the Single-cell Object

Here is the basic workflow that was used to make the single-cell object to use in the vignette. Notice there is a removal of BCR-related RNA features (using the Ibex function quietBCRgenes()). As we are going to combine multimodal data, both the GEX and CITE probes may cause bias in the weighted output.

##################################
#scRNA/ADT loading and processing
#################################
tmp <-  Read10X("~/data/GSM5073055_P1.1_filtered_feature_bc_matrix")

MIS.sample <- CreateSeuratObject(counts = tmp$`Gene Expression`)
rownames(tmp$`Antibody Capture`) <- stringr::str_remove_all(rownames(tmp$`Antibody Capture`), "anti_human_")
rownames(tmp$`Antibody Capture`) <- stringr::str_remove_all(rownames(tmp$`Antibody Capture`), "anti_mousehuman_")
rownames(tmp$`Antibody Capture`) <- substr(rownames(tmp$`Antibody Capture`), 6, nchar(rownames(tmp$`Antibody Capture`)))

adt_assay <- CreateAssayObject(counts = tmp$`Antibody Capture`)


MIS.sample[["ADT"]] <- adt_assay
MIS.sample <- subset(MIS.sample, subset = nFeature_RNA > 100) 
MIS.sample  <- RenameCells(object = MIS.sample , new.names = paste0("MIS.sample_", rownames(MIS.sample[[]])))
MIS.sample[["mito.genes"]] <- PercentageFeatureSet(MIS.sample, pattern = "^MT-")
    
#Filtering step
standev <- sd(log(MIS.sample$nFeature_RNA))*2.5 #cutting off above standard deviation of 2.5
mean <- mean(log(MIS.sample$nFeature_RNA))
cut <- round(exp(standev+mean))
MIS.sample <- subset(MIS.sample, subset = mito.genes < 10 & nFeature_RNA < cut)

#Processing and Adding Contig Info
contigs <- read.csv("~/data/GSM5073091_PBMC_P1.1_MIS-C_Severe_BCR_filtered_contig_annotations.csv.gz")
clones <- combineBCR(contigs, samples = "MIS.sample", removeNA = TRUE)
MIS.sample <- combineExpression(clones, MIS.sample, cloneCall="aa")

#Subset only B cells (by contigs)
MIS.sample$BCR.recoverd <- "No"
MIS.sample$BCR.recoverd[!is.na(MIS.sample$CTaa)] <- "Yes"
MIS.sample <- subset(MIS.sample, BCR.recoverd == "Yes")

#Processing RNA
DefaultAssay(MIS.sample) <- 'RNA'
MIS.sample <- NormalizeData(MIS.sample) %>% FindVariableFeatures() %>% 
  quietBCRgenes() %>% ScaleData() %>% RunPCA(verbose = FALSE)

#Processing ADT
DefaultAssay(MIS.sample) <- 'ADT'
VariableFeatures(MIS.sample) <- rownames(MIS.sample[["ADT"]])
MIS.sample <- NormalizeData(MIS.sample, normalization.method = 'CLR', margin = 2) %>% 
  ScaleData() %>% RunPCA(reduction.name = 'apca')

###################################
#Making Example Data Set for Trex
#################################
meta <- MIS.sample[[]]
meta <- meta[sample(nrow(meta), nrow(meta)*0.33),]
ibex_example <- subset(MIS.sample, cells = rownames(meta))
save(ibex_example, file = "ibex_example.rda", compress = "xz")

Loading the Data Object

For the purpose of the vignette, we will load the full object. The data example built into the package (ibex_example) is derived from randomly sampling cells from Patient 1 (see above).

SeuratObj <- readRDS(url("https://www.borch.dev/uploads/data/Ibex_FullExample.rds"))

Running Ibex

Ibex.matrix Function

Ibex has 2 major functions - the first being Ibex.matrix(), which is the backbone of the algorithm and returns the encoded values based on the selection of variables. Unlike runIbex() below, Ibex.matrix() does not filter the input for only B cells with attached BCR data. In addition, Ibex.matrix() is compatible with the list output from the combineBCR() function from the scRepertoire R package, while runIbex() must be performed on a single-cell object.

chains

“Heavy” for Ig Heavy Chain
“Light” for Ig Light Chain

method

“encoder” for a convolution neural network (CNN) based encoding.
“geometric” for a geometric transformation.

encoder.model

“VAE” for a variational autoencoder
“AE” for a traditional autoencoder

encoder.input

“AF” to use Atchley factors
“KF” to use Kidera factors
“both” to use both Atchley and Kidera factors
“OHE” for a One Hot Autoencoder

theta
If choosing the geometric transformation, what value of theta to use (default is pi)

ibex_vectors <- Ibex.matrix(SeuratObj, 
                            chains = "Light",
                            encoder.input = "OHE")

## [1] "Calculating the encoding values..."

qplot(data = as.data.frame(ibex_vectors), Ibex_2, Ibex_3) + theme_classic()

runIbex

Additionally, runIbex() can be used to append the Seurat or Single-cell Experiment object with the Ibex vectors and allow for further analysis. Importantly, runIbex() will remove single cells that do not have recovered BCR data in the metadata of the object.

SeuratObj <- runIbex(SeuratObj, 
                     chains = "Heavy",
                     encoder.input = "KF", 
                     reduction.name = "ibex.KF")

## [1] "Calculating the encoding values..."

Using Ibex Vectors

After runIbex(), we have the encoded values stored under “Ibex…”. Using the Ibex reduction stored in Seurat, we can calculate the nearest neighbor and shared nearest neighbor indexes and generate a UMAP.

#Generating UMAP from ibex Neighbors
SeuratObj <- RunUMAP(SeuratObj, 
                     reduction = "ibex.KF",
                     dims = 1:30,
                     reduction.name = 'ibex.umap', 
                     reduction.key = 'ibexUMAP_')

#ibex UMAP
plot1 <- DimPlot(SeuratObj, reduction = "ibex.umap") + NoLegend()
plot2 <- DimPlot(SeuratObj, group.by = "CTaa", reduction = "ibex.umap") + 
  scale_color_viridis(discrete = TRUE, option = "B") + 
  theme(plot.title = element_blank()) +
  NoLegend()

plot1 + plot2

We now can use this in a similar way as other single-cell modalities and calculate weighted nearest neighbor (WNN). To check out more on WNN, please read the Satija’s group paper. We will use the RNA, ADT protein levels, and ibex vectors for the WNN calculations.

SeuratObj <- FindMultiModalNeighbors(
                    SeuratObj, 
                    reduction.list = list("pca", "apca", "ibex.KF"), 
                    dims.list = list(1:30, 1:20, 1:30), 
                    modality.weight.name = "RNA.weight"
                  )
SeuratObj <- RunUMAP(SeuratObj, 
                     nn.name = "weighted.nn", 
                     reduction.name = "wnn.umap", 
                     reduction.key = "wnnUMAP_")
SeuratObj <- FindClusters(SeuratObj, 
                          graph.name = "wsnn", 
                          resolution = 0.6,
                          algorithm = 3, verbose = FALSE)

#WNN UMAP
plot3 <- DimPlot(SeuratObj, reduction = "wnn.umap")
plot4 <- DimPlot(SeuratObj, reduction = "wnn.umap", group.by = "CTaa") + 
  scale_color_viridis(discrete = TRUE, option = "B") + 
  theme(plot.title = element_blank()) +
  NoLegend()

plot3 + plot4

Comparing the outcome to just one modality

We can also look at the differences in the UMAP generated from RNA, ADT, or Ibex as individual components. Remember, the clusters that we are displaying in UMAP are based on clusters defined by the weighted nearest neighbors calculated above.

SeuratObj <- RunUMAP(SeuratObj, 
                     reduction = 'pca', 
                     dims = 1:30, 
                     assay = 'RNA', 
                     reduction.name = 'rna.umap', 
                     reduction.key = 'rnaUMAP_')

SeuratObj <- RunUMAP(SeuratObj, 
                     reduction = 'apca', 
                     dims = 1:20, 
                     assay = 'ADT', 
                     reduction.name = 'adt.umap', 
                     reduction.key = 'adtUMAP_')

plot5 <- DimPlot(SeuratObj, reduction = "rna.umap") + NoLegend()
plot6 <- DimPlot(SeuratObj, reduction = "adt.umap") + NoLegend()
plot7 <- DimPlot(SeuratObj, reduction = "ibex.umap") + NoLegend()

plot5 + plot6 + plot7

CoNGA Reduction

Recent work has proposed using representative cells for the characterization of clone and gene expression relationships. In order to generate these representative cells, either a mean expression across a clone or using the PCA dimensional space to identify a single cell that has the minimum euclidean distance across a clone.

In order to generate a single-cell object based on the CoNGA approach, Ibex offers the function CoNGAfy(). For method, select either “mean” or “dist” as described above. After performing CoNGAfy(), the user can use any of the above reduction strategies.

CoNGA.seurat <- CoNGAfy(SeuratObj, 
                         method = "dist")

CoNGA.seurat <- runIbex(CoNGA.seurat, 
                        encoder.input = "KF", 
                        reduction.name = "ibex.KF")

## [1] "Calculating the encoding values..."

CoNGA.seurat <- CoNGA.seurat %>%
                  FindNeighbors(reduction = "ibex.KF") %>%
                  FindClusters(algorithm = 3)

## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
## 
## Number of nodes: 1012
## Number of edges: 35549
## 
## Running smart local moving algorithm...
## Maximum modularity in 10 random starts: 0.5521
## Number of communities: 7
## Elapsed time: 0 seconds

CoNGA.seurat <- RunUMAP(CoNGA.seurat, 
                        reduction = "ibex.KF", 
                        dims = 1:20, 
                        reduction.name = 'ibex.umap', 
                        reduction.key = 'ibexUMAP_')

DimPlot(CoNGA.seurat, reduction = "ibex.umap") + NoLegend()

Compiled: May 04, 2024