v1.3.1: Set adaptive adjustment of minimum clustering threshold and singletons backtracking as optional

1. BCR Clustering Updates

1.1 Added min_depth_thre_adjustment parameter to data.BCR.clusters():

If 'TRUE', the minimal clustering threshold is increased by 1 for every 100,000 entries of input data for high efficiency.

1.2 Added singletons_backtrack parameter to data.BCR.clusters():

If 'TRUE', backtracking singleton clonotypes that weren't clustered but have multiple sequences in raw data.

v1.3.0: Productive Sequence Filtering, Cluster Depth Limits & Raw Data Backtracking

1. Data Preprocessing Enhancements

1.1 Added productive_only parameter to data.preprocess() function:

When set to True, filters raw data based on the 'productive' column annotation (requires raw data to contain this column).

1.2 Modified data.preprocess() function:

To include raw_index and raw_indices columns in processed data:
(1) raw_index: stores the index of representative sequence in raw data for each clonotype.
(2) raw_indices: stores indices of all sequences in raw data for each clonotype.

2. BCR Clustering Updates

2.1 Added max_depth_thre parameter to data.BCR.clusters():

To set maximum depth limit for clusters (default=1000), preventing excessively deep clusters that could hinder MSA.

2.2 Renamed cluster_thre parameter to min_depth_thre:

To better reflect its purpose of setting minimum depth threshold for clusters.

2.3 Implemented singletons.backtrack() function:

To handle backtracking of singleton clonotypes that weren't clustered but have multiple sequences in raw data.

3. New Functionality

3.1 Added new data.clusters.backtrack() function:

To utilize raw_indices from clusters to trace back to original data information.

This release focuses on improving data traceability and providing more control over clustering parameters, particularly for handling productive sequences and managing cluster depth constraints.

v1.2.1: Fixed bugs in fastBCR-p and public antibody prediction.

· Fixed bugs in paired data processing and light chain segregation, improving overall workflow robustness.
· The public antibody prediction module now fully supports all prediction models (including heavy chain models “cdrh” and “cdrh3”, as well as light chain models “cdrl” and “cdrl3”).

v1.2.0: Integrating light-chain refinements into heavy-chain clustering to optimize clonal family inference and supporting public antibody prediction and filtering.

fastBCR-p is a novel computational pipeline that incorporates public antibody filtering and strategic light-chain segregation, significantly improving both the accuracy and throughput of large-scale paired BCR clustering.
Existing heavy chain clone family inference methods all suffer from the issue of confounded light chain features. To address this, based on the heavy chain clustering results from fastBCR, fastBCR-p split sequences within heavy chain clusters that have different light chain VJ genes.

v1.1.3: Fixed a naming error of 'clonal.tree.generation()'

We have corrected the identifier naming error in the fasta generated by clonal.tree.generation(). Previously, we used 'clonotype_index' to name identifiers, where each 'clonotype_index' represents a clonotype composed of the same IGHV gene, IGHJ gene and junction amino acid sequence. However, when constructing the clonal tree, we used the DNA sequence of the raw data, so the identifier should be named using 'sequence_id'.

v1.1.2: Allow filtering of low-frequency sequences during data preprocessing

We updated the data.pro() function to allow filtering out low-frequency sequences (which may have arisen due to sequencing errors) during data preprocessing. Users can define the filtering threshold by adjusting 'count_filter_thre'. It defaults to 'NA' which means no filtering is performed.

v1.1.1: Increase gap opening penalty of MSA

To obtain more accurate multiple sequence alignment results, we increased the gap opening penalty for amino acid sequences from 10 (default) to 20, thereby minimizing the addition of gaps in clusters with equal-length sequences.

v1.1

We introduce an R-based computational pipeline meticulously designed for the efficient analysis of BCR repertoire sequencing data. The pipeline is grounded in fastBCR, a heuristic algorithm tailored for prompt clonal family inference. Its capabilities are further enriched through the integration of a comprehensive suite of essential modules. These modules encompass V/J gene usage statistics, distribution of conserved motifs, construction of phylogenetic trees, analysis of affinity maturation, diversity assessment, and the capability to query neutralizing antibodies (NAbs). Moreover, we offer methodologies for scrutinizing variations in clonal family compositions across diverse groups. This comprehensive pipeline advances the computational analyses of BCR repertoire, offering a convenient and effective approach for the systematic investigation and understanding of the B cell immune response.

v1.0

fastBCR is an efficient tool tailored for B cell receptor (BCR) clonal family inference, particularly from extensive BCR repertoire datasets. This tool contributes to advancing our comprehension of B cell activation and antibody-related investigations. Its primary objective is the prompt identification of B cell clonal families within vast BCR heavy chain sequences. To further bolster post-clustering analysis, fastBCR offers an array of functional modules. These include multiple sequence alignment (MSA), phylogenetic tree construction, somatic hypermutation (SHM) evaluation, and class switch analysis (CSR). Additionally, fastBCR incorporates a BCR simulation module that aids in generating variable B cell clonal lineages with distinct mutation rates. These modules enhance the applicability of fastBCR in BCR repertoire analysis and the exploration of antibodies, making it a valuable asset in immunological research.