ScRepLoading Process Configuration

Purpose

Load single-cell TCR-seq or scBCR-seq data from various formats into a scRepertoire-compatible object. This process reads VDJ (variable, diversity, joining) receptor contig data from multiple single-cell sequencing platforms and prepares it for integration with scRNA-seq data.

When to Use

• When analyzing scTCR-seq or scBCR-seq data alongside scRNA-seq
• Required for TCR/BCR clonotype analysis (CDR3 clustering, clone expansion, TESSA analysis)
• Enables integration of immune receptor information with single-cell expression data
• Supports multiple sequencing platforms: 10x Genomics, AIRR, BD, Dandelion, Immcantation, MiXCR, ParseBio, TRUST4, WAT3R, Omniscope

Important: This process is automatically enabled when your sample info file contains TCRData or BCRData columns.

Configuration Structure

Process Enablement

[ScRepLoading]
cache = true  # Enable caching (default: true)

Input Specification

[ScRepLoading.in]
# Type: file
# Required: yes
# Description: Sample metadata file (tab-delimited) with TCR/BCR data paths
metafile = "path/to/sample_info.txt"

Required input file columns:

• Sample: Unique identifier for each sample (required)
• TCRData (for TCR analysis): Directory path to scTCR-seq data
• BCRData (for BCR analysis): Directory path to scBCR-seq data
• Additional columns: Treated as sample metadata (optional)

Data format requirements:

• 10x Genomics: Directory containing filtered_contig_annotations.csv or all_contig_annotations.csv
• AIRR format: Directory containing airr_rearrangement.tsv
• BD platform: Directory containing Contigs_AIRR.tsv
• Dandelion: Directory containing all_contig_dandelion.tsv
• Immcantation: Directory containing _data.tsv or similar
• JSON: File with .json extension
• MiXCR: Directory containing clones.tsv
• ParseBio: Directory containing barcode_report.tsv
• TRUST4: Directory containing barcode_report.tsv
• WAT3R: Directory containing barcode_results.csv
• Omniscope: Directory containing .csv files

Path handling:

• If TCRData/BCRData specifies a directory: Process uses scRepertoire::loadContigs() directly
• If TCRData/BCRData specifies a file: Creates symbolic link to temp directory for processing
• When filename is not recognized by scRepertoire: Set envs.format explicitly

Environment Variables

[ScRepLoading.envs]
# type: choice - Data type to load (default: "auto")
# Options:
#   "TCR" - T cell receptor data
#   "BCR" - B cell receptor data
#   "auto" - Auto-detect from column names in sample info
# Note: If both TCRData and BCRData present, TCR selected by default
type = "auto"

# format: choice - Format of TCR/BCR data files (optional)
# Options: auto, 10X, AIRR, BD, Dandelion, Immcantation,
#          JSON, MiXCR, Omniscope, ParseBio, TRUST4, WAT3R
# If not provided, scRepertoire guesses from filename
format = "auto"

# combineTCR: json - Extra arguments for scRepertoire::combineTCR()
# See: https://rdrr.io/github/ncborcherding/scRepertoire/man/combinetcr
combineTCR = {"samples": true}

# combineBCR: json - Extra arguments for scRepertoire::combineBCR()
# See: https://rdrr.io/github/ncborcherding/scRepertoire/man/combinebcr
combineBCR = {"samples": true}

# exclude: auto or list - Columns to exclude from metadata (default: auto)
# auto = ["BCRData", "TCRData", "RNAData"]
# Can also be comma-separated string: "BCRData,TCRData,RNAData"
exclude = "auto"

# tmpdir: str - Temporary directory for symbolic links (default: "/tmp")
tmpdir = "/tmp"

Detailed combineTCR Parameters

[ScRepLoading.envs.combineTCR]
# samples: bool or list - Sample labels (default: true)
# true = use Sample column from metadata
# false = no sample grouping
# list = explicit sample labels
samples = true

# ID: str or null - Additional sample labeling (optional)
# Adds prefix to barcodes to prevent duplicate issues
ID = null

# removeNA: bool - Remove cells with missing chain values (default: false)
# true = filter out cells with NA in any chain
# false = include cells with 1 NA value (default)
removeNA = false

# removeMulti: bool - Remove cells with >2 chains (default: false)
# true = filter out multi-chain cells (>2 chains)
# false = include multi-chain cells (default)
removeMulti = false

# filterMulti: bool - Select highest-expression chain for multi-chain (TCR default: false)
# true = keep highest UMI count chain if multiple chains present
# false = keep all chains (default)
filterMulti = false

# filterNonproductive: bool - Remove non-productive rearrangements (default: true)
# true = filter out non-functional receptors
# false = include all rearrangements
filterNonproductive = true

Detailed combineBCR Parameters

[ScRepLoading.envs.combineBCR]
# samples: bool or list - Sample labels (default: true)
samples = true

# ID: str or null - Additional sample labeling (optional)
ID = null

# call.related.clones: bool - Cluster related BCR clones (default: true)
# Uses nucleotide sequence + V gene with Levenshtein distance
# false = uses V gene + amino acid sequence for CTstrict
call.related.clones = true

# threshold: num - Normalized edit distance for clustering (default: 0.85)
# Higher = more permissive clustering (more sequences grouped)
# Range: 0.0 - 1.0
threshold = 0.85

# removeNA: bool - Remove cells with missing chain values (default: false)
removeNA = false

# removeMulti: bool - Remove cells with >2 chains (default: false)
removeMulti = false

# filterMulti: bool - Select highest-expression chain (default: true)
# true = keep highest UMI count chain
# false = keep all chains
filterMulti = true

# filterNonproductive: bool - Remove non-productive rearrangements (default: true)
filterNonproductive = true

Configuration Examples

Minimal Configuration (10x TCR Data)

[SampleInfo.in]
infile = "sample_info.txt"

# Sample info file contents:
# Sample  Age  Sex  Diagnosis  RNAData            TCRData
# C1      62   F    Colitis    /data/C1/rna      /data/C1/tcr
# C2      71   F    Colitis    /data/C2/rna      /data/C2/tcr

# ScRepLoading auto-enables when TCRData column present
# No explicit ScRepLoading section needed

Single Sample with Format Specification

[ScRepLoading]
cache = true

[ScRepLoading.in]
metafile = "metadata/single_sample.txt"

[ScRepLoading.envs]
type = "TCR"
format = "10X"

[ScRepLoading.envs.combineTCR]
removeNA = true
filterNonproductive = true

Multi-Sample BCR Analysis with Clustering

[ScRepLoading]
cache = true

[ScRepLoading.in]
metafile = "metadata/bcr_samples.txt"

[ScRepLoading.envs]
type = "BCR"

[ScRepLoading.envs.combineBCR]
call.related.clones = true
threshold = 0.85  # Higher threshold for more permissive clustering
filterMulti = true
removeMulti = false

Non-10x Format (AIRR)

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/airr_samples.txt"

[ScRepLoading.envs]
format = "AIRR"
type = "auto"

[ScRepLoading.envs.combineTCR]
removeNA = false
removeMulti = false

TRUST4 Format

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/trust4_samples.txt"

[ScRepLoading.envs]
format = "TRUST4"

[ScRepLoading.envs.combineTCR]
removeNA = true
filterNonproductive = true

Common Patterns

Pattern 1: 10x Genomics TCR Data (Most Common)

# sample_info.txt
# Sample    RNAData              TCRData
# Sample1   /data/Sample1/rna   /data/Sample1/vdj
# Sample2   /data/Sample2/rna   /data/Sample2/vdj

[SampleInfo.in]
infile = "sample_info.txt"

# TCR directories must contain filtered_contig_annotations.csv
# No ScRepLoading configuration needed - auto-detected

Pattern 2: Both TCR and BCR Data (Auto-Detect TCR)

# sample_info.txt
# Sample    RNAData              TCRData              BCRData
# Sample1   /data/Sample1/rna   /data/Sample1/tcr   /data/Sample1/bcr

[SampleInfo.in]
infile = "sample_info.txt"

# TCR selected by default when both columns present
# To explicitly analyze BCR instead:
[ScRepLoading.envs]
type = "BCR"

Pattern 3: Filtered TCR Data (Remove NA and Multi-Chain)

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/tcr_filtered.txt"

[ScRepLoading.envs.combineTCR]
removeNA = true      # Remove cells with missing chains
removeMulti = true   # Remove cells with >2 chains
filterNonproductive = true  # Remove non-functional receptors

Pattern 4: Relaxed Filtering for Exploratory Analysis

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/tcr_exploratory.txt"

[ScRepLoading.envs.combineTCR]
removeNA = false     # Keep cells with single chain
removeMulti = false  # Include multi-chain cells for inspection
filterNonproductive = false  # Include non-productive rearrangements

Pattern 5: BCR Clone Clustering with Custom Threshold

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/bcr_clustering.txt"

[ScRepLoading.envs.combineBCR]
call.related.clones = true
threshold = 0.90  # More stringent clustering (lower = more permissive)

Pattern 6: Sample-Specific Labeling

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/longitudinal.txt"

[ScRepLoading.envs.combineTCR]
samples = true  # Use Sample column from metadata
ID = "Timepoint"  # Add Timepoint as additional label prefix

# Creates barcodes like: "Sample1_Timepoint1_AAACCC..."
# Prevents duplicate barcode issues across timepoints

Pattern 7: Custom Metadata Exclusion

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/custom_columns.txt"

[ScRepLoading.envs]
exclude = ["RNAData", "TCRData", "BCRData", "ExperimentID", "Batch"]

# These columns excluded from scRepertoire object metadata
# Helps reduce metadata clutter in downstream analysis

Pattern 8: Paired Chain Analysis (TRA+TRB for TCR)

# Default behavior - ScRepLoading automatically pairs chains
# at cell barcode level when both TRA and TRB present

[ScRepLoading]

[ScRepLoading.in]
metafile = "metadata/tcr_paired.txt"

[ScRepLoading.envs.combineTCR]
removeNA = false  # Keep single-chain cells for inspection
filterMulti = false  # Don't filter multi-chain cells

# Later analysis can filter for true paired chains
# Using downstream processes like CDR3Clustering

Dependencies

Upstream Processes

• SampleInfo (required): Provides sample metadata with TCRData/BCRData columns
• LoadingRNAFromSeurat (alternative): When loading RNA from Seurat instead of SampleInfo

Downstream Processes

• ScRepCombiningExpression: Integrates TCR/BCR data with scRNA-seq expression
• CDR3Clustering: Clones cells by CDR3 sequence similarity
• TESSA: TCR-specific analysis (epitope specificity prediction)
• CDR3AAPhyschem: Physicochemical properties of CDR3 sequences
• ClonalStats: Clonality statistics and diversity metrics

Validation Rules

Common Configuration Errors

1. Missing TCRData/BCRData column:

   • Error: Process not enabled, no TCR/BCR analysis
   • Fix: Add TCRData or BCRData column to sample info file

2. Invalid format specified:

   • Error: scRepertoire fails to recognize file format
   • Fix: Set envs.format to one of: 10X, AIRR, BD, Dandelion, Immcantation, JSON, MiXCR, ParseBio, TRUST4, WAT3R, Omniscope

3. Directory path not found:

   • Error: Cannot access TCR/BCR data directory
   • Fix: Verify paths in TCRData/BCRData columns exist and are readable

4. Missing required files in directory:

   • Error: Expected contig file not found (e.g., filtered_contig_annotations.csv)
   • Fix: Ensure directory contains appropriate file for specified format

5. Both TCR and BCR specified without type selection:

   • Warning: TCR selected by default
   • Fix: Set envs.type = "BCR" if BCR analysis intended

File Format Requirements

• 10x Genomics: Must have filtered_contig_annotations.csv in directory
• AIRR: Must have airr_rearrangement.tsv in directory
• BD: Must have Contigs_AIRR.tsv in directory
• Dandelion: Must have all_contig_dandelion.tsv in directory
• MiXCR: Must have clones.tsv in directory
• TRUST4: Must have barcode_report.tsv in directory
• ParseBio: Must have barcode_report.tsv in directory
• WAT3R: Must have barcode_results.csv in directory

Chain Compatibility

• TCR chains: Supports TRA, TRB, TRG, TRD (auto-detected from data)
• BCR chains: Supports IGH, IGL, IGK (auto-detected from data)
• Paired analysis: Automatically pairs TRA+TRB or IGH+IGL/IGK when both present
• Single-chain: Keeps single-chain cells when removeNA = false

Troubleshooting

Issue: ScRepLoading not running

Cause: No TCRData or BCRData column in sample info file
Solution:

1. Add TCRData or BCRData column to sample info
2. Verify column name exactly matches (case-sensitive)
3. Check that SampleInfo.in.infile is correctly specified

Issue: "File format not recognized"

Cause: Filename doesn't match expected pattern for auto-detection
Solution:

1. Set envs.format explicitly to your format type
2. Example: format = "TRUST4" for TRUST4 output
3. Verify directory contains expected file for that format

Issue: "No cells loaded" or empty output

Cause: Too aggressive filtering or mismatched barcodes
Solution:

1. Set removeNA = false and removeMulti = false temporarily
2. Check that TCR/BCR barcodes match RNA barcodes
3. Verify filterMulti is appropriate for your data type

Issue: Duplicate barcode errors

Cause: Multiple samples have identical cell barcodes
Solution:

1. Set ID = "Sample" or use explicit sample labels
2. This adds sample prefix to barcodes: Sample1_AAACCC...
3. Required when merging samples from same run

Issue: BCR clustering too strict/too permissive

Cause: Default threshold (0.85) not optimal for data
Solution:

1. Adjust envs.combineBCR.threshold
2. Higher (0.90+): More stringent, fewer clusters
3. Lower (0.80-): More permissive, more sequences clustered together

Issue: Single-chain cells lost

Cause: filterNonproductive = true or removeNA = true
Solution:

1. For exploratory analysis, set removeNA = false
2. For developmental studies, consider filterNonproductive = false
3. Use filterMulti = true only when confident in data quality

Issue: Metadata columns missing from output

Cause: Excluded by default (exclude = "auto")
Solution:

1. Set exclude = [] to keep all metadata columns
2. Or specify custom list: exclude = ["RNAData"]
3. Default excludes: RNAData, TCRData, BCRData

Issue: Cannot load from specific directory path

Cause: Path not accessible or permission issues
Solution:

1. Verify directory exists and is readable
2. Check file permissions: ls -la path/to/tcr/
3. Use absolute paths if relative paths fail

Issue: Combining TCR and BCR data separately

Cause: Need to analyze both receptor types
Solution:

1. Run pipeline twice with different type settings
2. First run: [ScRepLoading.envs] type = "TCR"
3. Second run: [ScRepLoading.envs] type = "BCR"
4. Use different output directories to avoid conflicts

Issue: Integration with ScRepCombiningExpression fails

Cause: Barcodes don't match between RNA and VDJ data
Solution:

1. Ensure same samples used in both RNA and VDJ data
2. Check that SampleInfo has correct paths for both data types
3. Verify barcode prefixes match (if using ID parameter)