Lintind

LinTInd

Workflows

Standard Workflow

Reconstruct a lineage tree and visualize indel patterns from CRISPR-edited amplicon sequencing data.

library(LinTInd)

# Define paths to package raw data
data_path <- paste0(system.file("extdata", package = 'LinTInd'), "/CB_UMI")
fafile <- paste0(system.file("extdata", package = 'LinTInd'), "/V3.fasta")
cutsite_path <- paste0(system.file("extdata", package = 'LinTInd'), "/V3.cutSites")
celltype_path <- paste0(system.file("extdata", package = 'LinTInd'), "/celltype.tsv")

# Read input files
data <- read.table(data_path, sep="\t", header=TRUE)
ref <- ReadFasta(fafile)
cutsite <- read.table(cutsite_path, col.names = c("indx", "start", "end"))
celltype <- read.table(celltype_path, header=TRUE, stringsAsFactors=FALSE)

# Align reads and identify indels
scarinfo <- FindIndel(data=data, scarfull=ref, scar=cutsite, indel.coverage="All", type="test", cln=1)

# Generate array-form scar strings
scarinfo <- IndelForm(scarinfo, cln=1)

# Define final scar form for each cell
cellsinfo <- IndelIdents(scarinfo, method.use="umi.num", cln=1)

# Visualize indel distribution
IndelPlot(cellsinfo = cellsinfo)

# Extract indels and calculate group similarity
tag <- TagProcess(cellsinfo$info, Cells=celltype)
tag_dist <- TagDist(tag, method = "Jaccard")

# Reconstruct and plot the lineage tree
treeinfo <- BuildTree(tag)
plotinfo <- PlotTree(treeinfo = treeinfo, data.extract = "TRUE", annotation = "TRUE")
plotinfo$p

Note on inputs/outputs:

• Input: A data frame of sequencing reads (with optional cell barcodes and UMIs), a reference FASTA file, a cut site position table, and an optional cell type annotation table.
• Output: A reconstructed lineage tree object and visualization plots of indels and tree structure.

When to Use

• CRISPR Lineage Tracing: Reconstructing lineage trees from alleles generated by CRISPR-mediated gene editing using BuildTree().
• Indel Identification: Aligning sequencing reads to a reference sequence and identifying insertions/deletions using FindIndel().
• Scar Profiling: Generating standardized array-form scar strings for each read or single cell using IndelForm() and IndelIdents().
• Group Similarity Analysis: Quantifying similarity (e.g., Jaccard, Spearman, or hypergeometric test) between cell groups/clusters using TagDist().

When NOT to Use

• General Single-Cell Analysis: For standard single-cell RNA-seq clustering or cell-type annotation, use Seurat or scran because LinTInd is specialized for lineage tracing via CRISPR-induced scars.
• Phylogenetics from SNPs: For standard phylogenetic tree reconstruction from natural genomic mutations (e.g., SNPs) without CRISPR-induced scars, use ape because LinTInd relies on specific cut-site and indel array structures.

Data Requirements

• Sequence Data: A data frame containing a column of read sequences (Read.Seq), with optional columns for cell barcodes (Cell.BC) and UMIs (UMI).
• Reference Sequence: A FASTA file containing the reference sequence, loaded using ReadFasta().
• Cut Sites: A table containing cut site indices, start positions, and end positions.
• Cell Type Annotation (optional): A tab-separated file mapping cell barcodes (Cell.BC) to cell types (Cell.type).

Key Parameters

• indel.coverage ("All"): Parameter in FindIndel() specifying which indels to cover.
• type ("test"): Parameter in FindIndel() specifying the run type.
• cln (1): Number of cores/threads to use in FindIndel(), IndelForm(), and IndelIdents().
• method.use ("reads.num"): Method to define the scar form per cell in IndelIdents(); options include "reads.num", "umi.num", and "consensus".
• method ("Jaccard"): Similarity calculation method in TagDist(); options include "Jaccard", "P", and "spearman".
• data.extract ("TRUE"): Logical parameter in PlotTree() to extract data.
• annotation ("TRUE"): Logical parameter in PlotTree() to annotate the tree plot.

Best Practices

• Verify that the reference sequence and cut sites match the experimental design before running FindIndel().
• Use IndelForm() immediately after FindIndel() to convert identified indels into standardized array-form strings.
• For single-cell data with UMIs, use method.use="umi.num" in IndelIdents() to define cell scar forms based on UMI consensus rather than raw read counts.
• Provide cell type annotations to TagProcess() to enable group-level similarity calculations with TagDist().

Common Pitfalls

• Missing Sequence Columns: Missing or mismatched column names in the input sequence data frame (e.g., missing Read.Seq). Fix: Ensure the sequence column is present and correctly named before running FindIndel().
• Missing Annotations in TagDist: Running TagDist() without cell type annotations. Fix: Pass a valid cell type data frame to TagProcess() via the Cells parameter before computing distances.
• High Computational Time: High computational time when processing large datasets. Fix: Increase the cln parameter in FindIndel(), IndelForm(), and IndelIdents() to utilize multiple cores.

Alternatives

• Seurat for general single-cell analysis and clustering.
• scater for single-cell quality control and visualization.
• scran for single-cell expression data normalization and variance modeling.
• ape for general phylogenetic tree reconstruction and analysis.

Citations

• Wang L. (2021). LinTInd: Reconstruction of lineage trees from CRISPR-mediated gene editing. R package.

References

• Homepage: https://bioconductor.org/packages/lintind
• Vignette: https://bioconductor.org/packages/release/bioc/vignettes/lintind/inst/doc/vignette_main.html