Ajaingithub/scCITESeq_pipeline

Batch and Modality Integration of single cell CITEseq dataset

scCITESeq

The functions can be used to perform the downstream analysis of the scCITESeq data. I have used all these function in pipeline_functions/scCITESeq_Execution.R

1. scCITESeq_QC function - This function will perform the quality check scRNA and scADT dataset before and after fitering the cells based on the cell number, number of genes, mitochondir percent, and ADT UMI.

   To Run the Code

   source("./pipeline_functions/CR7_scCITESeq_QC.R")
   
   CR_outdir ="./post_processing/genotype_demux/Exp01_geno/outs/per_sample_outs/Exp01_geno"
   
   scCITE_QC(samplepath = CR_outdir,
             samplename = "BB22003",
             saveDir = "./output/",
             min_cells = 3,
             min_genes = 200,
             max_genes = 5000,
             mitopercent = 10,
             ADT_UMI = 100000)
   

   This will create the filtered RNA and ADT (protein) object

2. scCITEseq_Doublet_Finder - This function will remove the doublets from the object.

   To Run the Code

      source("./pipeline_functions/scCITEseq_Doublet_Finder.R")
   
      doublet_scCITEseq(Obj= RNA_obj,
                        dims = 30,
                        res = 0.8,
                        saveDir = "./output/",
                        Obj2 = ADT_obj,
                        samplename = "BB22003",
                        process = "Doublet_finder",
                        Assay = "RNA")
   

   This will filter out doublets from both the RNA and ADT

   

3. scCITESeq_sctransform_V2 - The object is normalized using scTransform V2 and samples integreated using CCA.

   To Run the Code

      source("./pipeline_functions/scCITESeq_sctransform_V2.R")
   
      sctransform_V2_integration(obj = RNA_obj,
                                 saveDir = "./output/" ,
                                 ngenes=4000,
                                 regress = "mito_score",
                                 Assay = "RNA",
                                 process ="norm",
                                 objname = "BB22003",
                                 split_by="orig.ident",
                                 reference = NULL,
                                 dims=30,
                                 sample_tree=NULL,
                                 k_weight=100)
   

4. scCITESeq_RNA_intergation - The RNA / ADT object integration function.

   To Run the Code

     source("pipeline_functions/scCITESeq_RNA_intergation.R")
   
     RNA_integration(obj = RNA_obj/ADT_obj,
                     saveDir = "./output/",
                     dims=30,
                     RNA_features = c("CD4","CD8A"), 
                     Assay="RNA",
                     process="integration",
                     objname = "BB22003",
                     ncol=2,
                     ndims = 50)
   

5. scCITESeq_ADT_merging - Performing the ADT (protein) Sample Integration using CCA.

   To Run the Code

         source("pipeline_functions/scCITESeq_ADT_merging.R")
   
         ADT_merging(ADT_obj = ADT_obj,
                     saveDir= "./output/",
                     dims=10,
                     numfeatures=NULL,
                     Assay="ADT",
                     process="ADT_integration",
                     objname="BB22003",
                     split_by="orig.ident", 
                     reference=NULL,
                     sample_tree = NULL,
                     k_weight=100)
   

   

6. cluster_UMAP_QC - Performing clustering and checking the quality for each cluster.

   To Run the Code

      source("pipeline_functions/cluster_UMAP_QC.R")
   
      cluster_UMAP_and_QC(obj_path=RNA_obj/ADT_obj,
                          dims = 30,
                          res = 0.8,
                          saveDir = "./output/",
                          Assay = "RNA",
                          QC_features = c("nCount_RNA","nFeature_RNA"),
                          objname = "BB22003",
                          process = "UMAP_QC",
                          col_sel = c("Age","Run","orig.ident","gender"))
   

   

7. scCITEseq_modality_integration - Modality RNA and ADT integration.

   To Run the Code

          source("pipeline_functions/scCITEseq_modality_integration.R")
   
          modality_integration(RNA_obj = RNA_obj,
                               ADT_obj = ADT_obj,
                               RNA_dims = 30,
                               ADT_dims = 10,
                               saveDir = savedir,
                               Assay = "integrated",
                               process = "RNA_ADT",
                               objname= "BB22003")
   

These 7 step will perform the batch as well as modality (RNA and protein) integration for the scCITESeq experiment

Downstream Analysis

Pseudobulk PCA and Differential:

       source("pipeline_functions/Psuedobulk_PCA.R")
       
       CD8_subset_dds <- COVID_pseudobulk_within_cluster_AJ(obj = CD8_mem, 
                                                           savedir = savedir, 
                                                           group1 = group1, 
                                                           group2 = group2,
                                                           grouping_by = "Age", 
                                                           cluster = "all", 
                                                           cell_freq = 20, 
                                                           remove_samples = remove_samples,
                                                           cluster_group = "seurat_clusters",
                                                           sample_col = "orig.ident", 
                                                           batch_col = "Run",
                                                           gene_min_counts =  5, 
                                                           column_name_split = c("sampleid","virus","age","age_number","gender","Run"))
                                                           


     source("pipeline_functions/Pseudobulk_differential.R")
     
     design0 <- model.matrix(~ 0 + Age + Run, data = colData(CD8_subset_dds))
     colnames(design0) <- c("O","Y",paste("Run",1:(ncol(design0)-2),sep = ""))
     cm <- makeContrasts(O_VS_Y = O-Y,levels = design0)
     
     desl_clus <- LimmaEdgeR_differential(dds = CD8_subset_dds,
                                          design0 = design0,
                                          cm = cm, 
                                          savedir = savedir2,
                                          logfc = 0.5,
                                          p_value_adj = 0.05)