BD Rhapsody™ Sequence Analysis Pipeline

v2.1 BD Rhapsody™ Sequence Analysis Pipeline

New additions

• TCR/BCR high-quality cell designation and associated metrics. This creates a new set of VDJ metrics similar to products where there is a putative cell call for VDJ libraries, separate from the cell call from associated gene expression libraries
  
• UMAP dimensionality reduction coordinates as an output file and also builds those coordinates into the pipeline report, Seurat and Scanpy outputs
  
• Extra utility for only annotating the cell index and UMI of R1 and putting it in the header of R2

Updates

• Seurat output to separate mRNA and AbSeq data into RNA and ADT assays, respectively
  
• Scanpy output to use Muon (.h5mu) and create mRNA and AbSeq data in separate anndata objects, rna and prot respectively
  
• TCR/BCR dominant contigs file to include AIRR compliant germline columns and to only retain cell type appropriate chains. All chains are still available in the unfiltered contigs file.
  
• TCR/BCR dominant contigs file to rename the column 'duplicate_count' to 'umi_count', in accordance with AIRR's definition update in v1.4.1
  
• TCR/BCR dominant contig selection process, elevating the importance of a productive contig with high relative read count and removing the CDR3 requirement
  
• TCR/BCR DBEC algorithm to allow exceptions for CDR3 sequences not seen in any other cell and CDR3 paired chains seen in other cells
  
• TCR/BCR contig_id to correspond with annotated chain type
  
• Basic cell calling to scale better with small and large cell datasets and prevent most inappropriately high-cell calls derived from noise signatures
  
• Alignment Category 'No_Feature_Pct' metric to include targeted mRNA reads that are filtered out due to an invalid alignment
  
• Cell label annotation to improve the speed of annotation for reads with cell label sequences that contain more than 1 error
  
• RAM requirements for VDJ_preprocess_reads on local server runs
  
• Error handling and reporting in read processing steps
  
• Logging to capture errors during alignment with STAR
  
• FASTQ handling to skip reads with empty sequence
  
• Cell type classification model selection to better select an appropriate model when not all bioproducts are found in any one model
  
• Pipeline report to show sub-sampled tSNE and UMAP plots in the case where the putative cell count exceeds 100,000 and to show details of refined cell calling when refined cell calling is selected
  
• Bead version detection and read trimming

Fixes

• Issue that caused failure when a gene symbol was named 'nan'
  
• Issue with a quote mark in a gene symbol causing a failure in the Seurat output file generation
  
• Rare division by zero issue in DBEC algorithm
  
• Rare issue caused by including "SampleTag" in the Run_Name parameter

Experimental

• Added docker-free version of the pipeline, available for local server installs as a tar.gz bundle. Tested on Linux versions: Ubuntu 16 / 20 / 22 - Red Hat 7 - CentOS 7 / 9

make_rhapsody_reference tool:

• Added an 'Extra_STAR_params' input to enable passing parameters to the STAR genomeGenerate process
  
• Updated to automatically generate a GTF for sequences added in the 'Extra_sequences' FASTA input—useful for transgenes