IgBLAST

IgBLAST online overview

A heavy-chain read from an antibody repertoire is about 360 nucleotides of rearranged DNA. IgBLAST reports which V, D and J germline genes it came from, how far the V region has mutated away from germline, where the CDR3 sits, and whether the rearrangement is productive. This page runs the NCBI IgBLAST tool, version 1.22.0, online against IMGT germline references for human, mouse, rabbit, rat and rhesus monkey immunoglobulins, and human and mouse T cell receptors.

Nucleotide queries run igblastn and return the full AIRR rearrangement table, a clonotype report and the native files. Protein queries run igblastp, which assigns the V gene and V domain boundaries only. Each run is saved to the job history with its results table and downloads.

IgBLAST was published by Jian Ye, Ning Ma, Thomas Madden and James Ostell at NCBI in 2013. It is used to annotate antibodies from discovery campaigns, track V gene usage and CDR3 lineages after vaccination, profile B cell repertoires in autoimmunity and cancer, and pick germline templates for antibody design and humanization.

Pricing

Jobs start at 5 credits. The quote is calculated before submission from the total sequence letters across every record in the job: nucleotides and amino acids both count, while FASTA headers, whitespace and gap characters (- and .) do not.

Sequences in one job	Protein VH, 120 aa each	Heavy-chain nucleotide, 360 nt each	Light-chain nucleotide, 330 nt each
1	5	5	5
10	8	13	12
100	26	56	52
1,000	132	297	279

These cells come from the current calculator with FASTA input of identical records. Organism, receptor type, domain system, alignment penalties and the number of reported alignments do not change the quote.

Inputs

Input	Accepted formats	Limits
Antibody/TCR sequences	Pasted text, or a .fasta, .fa, .fas or .txt file.	One 50 MiB input. Free plans accept up to 10 sequences and 10,000 sequence letters per job.

Paste one raw sequence or any number of FASTA records. A raw sequence without a header is named query_1. Records may be DNA, RNA or protein; with Sequence type set to Auto-detect, a query whose informative characters are all IUPAC nucleotide codes runs as nucleotide, and anything else runs as protein. Nucleotide input gives the full V(D)J analysis.

Settings

Sequence configuration

Parameter	Type	Default	Description
Organism	enum	Human	Germline reference set: Human, Mouse, Rabbit, Rat or Rhesus monkey.
Receptor type	enum	Ig	Immunoglobulin (antibody) or T cell receptor. TCR runs accept human and mouse only.
Sequence type	enum	Auto-detect	Auto-detect, Nucleotide (DNA/RNA) or Protein (amino acid). Protein runs assign the V gene only.

Germline annotation

Parameter	Type	Default	Description
Domain system	enum	IMGT	Framework and CDR delineation scheme (-domain_system): IMGT or Kabat.

Every run uses checksum-verified IMGT germline references from the 2026-07-26 snapshot, prepared with the IgBLAST release's own edit_imgt_file.pl. The snapshot date is recorded in the result metadata.

Nucleotide alignment

These options map to igblastn arguments. A protein run with any of them changed from its default is rejected before it starts, because igblastp does not accept them.

Parameter	Type	Default	Description
Query strand	enum	Both strands	Strands searched (-strand): Both strands, Plus strand or Minus strand.
V mismatch penalty	integer	-1	V gene mismatch penalty (-V_penalty), from -3 to -1.
D mismatch penalty	integer	-2	D gene mismatch penalty (-D_penalty), from -4 to -1.
J mismatch penalty	integer	-2	J gene mismatch penalty (-J_penalty), from -3 to -1.
Minimum V gene length	integer	9	Minimum V alignment length (-min_V_length), at least 9.
Minimum J gene length	integer	0	Minimum J alignment length (-min_J_length), at least 0.
Extend V alignment at 5' end	boolean	false	Extends the V alignment to the query's 5' end (-extend_align5end).
Extend J alignment at 3' end	boolean	false	Extends the J alignment to the query's 3' end (-extend_align3end).
Allow V(D)J overlap	boolean	false	Lets V, D and J alignments overlap (-allow_vdj_overlap). IgBLAST applies it only with D penalty -4 and J penalty -3.

Output options

Parameter	Type	Default	Description
V gene alignments	integer	3	V gene matches reported per query (-num_alignments_V), 1 to 10.
D gene alignments	integer	3	D gene matches reported per nucleotide query (-num_alignments_D), 1 to 10.
J gene alignments	integer	3	J gene matches reported per nucleotide query (-num_alignments_J), 1 to 10.
Max clonotypes	integer	100	Top clonotypes written to the nucleotide clonotype report (-num_clonotype), 0 to 1,000. 0 lists none.

Outputs

The Results table has one row per query and can be copied or downloaded. Nucleotide rows come from the AIRR rearrangement table; protein rows show the top V assignment from the native format 7 report.

Column	Description
Query ID	Identifier from the FASTA header, or query_1 for a raw sequence.
Chain	VH, VL-KAPPA or VL-LAMBDA for antibodies; VA, VB, VG or VD for TCRs. The native AIRR locus is kept in the files.
V Gene	Best V allele, for example IGHV3-23*01. Several comma-separated alleles mean the scores tied.
V Identity %	Percent identity between the query and the assigned V germline allele over the aligned region.
D Gene	Best D allele for heavy chains and TCR beta or delta. N/A for light chains and protein runs.
J Gene	Best J allele. N/A for protein runs.
CDR3 (AA)	CDR3 amino acid sequence.
CDR3 (NT)	CDR3 nucleotide sequence.
Junction	Junction nucleotides, which add the conserved cysteine and tryptophan or phenylalanine codons on either side of CDR3.
Productive	Yes when the rearrangement is in frame without a stop codon, No otherwise.

File	Contents
igblast-airr.tsv	Nucleotide runs: the complete AIRR rearrangement table, with alignments, region boundaries, scores and supports.
igblast-format7.txt	Protein runs: the native format 7 report with the FR1 to CDR3 domain table and every V hit.
igblast-clonotypes.txt	Nucleotide runs with more than one query: IgBLAST's summary of the top clonotypes in the batch.
igblast.log	Command, return code and IgBLAST messages.
native/ files	The unmodified IgBLAST output, prepared query, stdout and stderr, and scientific-data.json with the command, settings and reference checksums.

Understanding results

V identity

V identity measures how far the V region has drifted from its closest germline allele, which mostly reflects somatic hypermutation. As a rough guide, 95 to 100% is germline-like and typical of naive B cells, 85 to 95% suggests a moderately mutated antibody from an early response, and values below 85% point to extensive affinity maturation in memory B cells.

Identity is measured against the IMGT reference set, not the donor's own genome. A donor carrying an allele missing from the reference will show a few apparent mutations in every sequence from that gene, so a constant low-level offset across a repertoire is worth checking against other genes before reading it as hypermutation.

CDR3 and junction

CDR3 spans the V-D-J joint, so it carries most of the junctional diversity and usually most of the antigen contact. Sequences that share V gene, J gene and CDR3 length, with near-identical CDR3s, most likely descend from one B cell; this is how clonal lineages are grouped, and the clonotype report summarizes the most common clonotypes in a nucleotide batch. The junction is CDR3 plus the two conserved anchor codons on either side.

D gene calls

D genes are short and lose bases at both ends during recombination, while untemplated nucleotides are added around them. A heavy chain often matches several D alleles equally well, for example IGHD3-10*01,IGHD3-10*02,IGHD3-16*01. The first allele is the top-scoring call. Treat ambiguous D calls as uncertain and rely on V, J and CDR3 for lineage grouping.

Productive and protein results

A No in Productive marks an out-of-frame junction or a stop codon. In repertoire data this usually comes from nonproductive rearrangements on the second allele, sequencing errors or PCR artifacts, not from expressed antibodies. Protein runs cannot see junctions or reading frames, so they report the V gene and V domain boundaries only; use nucleotide input when D, J, CDR3 and productivity matter.

To renumber the variable domain in Chothia, Martin or AHo, run ANARCI on the same sequences. BioPhi scores humanness against human repertoires, and ABodyBuilder3 predicts the Fv structure from paired heavy and light chains.

ProteinIQ and NCBI IgBLAST

NCBI hosts the official IgBLAST web server with separate immunoglobulin and T cell receptor forms. ProteinIQ is not an NCBI service. It runs the same IgBLAST 1.22.0 release from NCBI on its own infrastructure, so queries are not sent to NCBI.

The NCBI form offers more choices for the search itself: a choice of IMGT, AIRR-C, NCBI and KIMDB germline sets or a custom database, constant-region annotation against C gene databases, GenBank accession or GI numbers as queries, searches of additional databases such as nt and RefSeq, and query-anchored alignment views. ProteinIQ uses the IMGT snapshot only and offers none of those options, nor the Expect and minimum D match settings.

What ProteinIQ adds is around the run. Each job is saved with its inputs, settings, results table and the native AIRR, clonotype and log files, so a repertoire can be reopened, shared or downloaded later. Paid plans can also submit IgBLAST jobs through the ProteinIQ API.

How IgBLAST works

IgBLAST is not a single BLAST search. It searches the query against the V gene database first, using parameters tuned for a segment of about 300 bases. It then masks the matched V region and searches the rest of the query against the D and J databases with settings suited to short segments. Each segment has its own mismatch penalty, the values listed under Settings.

Each call must also make biological sense. The D gene has to fall between V and J, all segments must come from the same locus (IGH, IGK or IGL for antibodies; TRA, TRB, TRG or TRD for TCRs), and light chains are read as V-J joins without a D. Framework and CDR boundaries are transferred from the annotated germline V gene to the query through the alignment, so insertions and deletions in the query shift the boundaries correctly.

V(D)J recombination background

Each B and T cell builds its receptor gene by cutting and joining segments from arrays in the germline DNA. A heavy chain joins one V, one D and one J segment; light chains join only V and J. The human heavy-chain locus holds roughly 40 to 50 functional V genes, 23 functional D genes and 6 functional J genes, which is where combinatorial diversity comes from.

The joins add more diversity than the combinations. Exonucleases trim bases from the segment ends, terminal deoxynucleotidyl transferase adds random N nucleotides, and hairpin opening can add short palindromic P nucleotides. All of this lands in CDR3, which is why IgBLAST annotates the junction separately.

After antigen activation, somatic hypermutation in germinal centers adds point mutations across the V region, and selection for tighter binders drives affinity maturation. The V identity in the results is the trace that process leaves.

Limitations

Gene calls are only as good as the germline reference. Species outside the five offered, and individuals with alleles not yet in IMGT, get approximate assignments, and custom germline databases are not available here. TCR analysis is limited to human and mouse.

D gene assignment is often ambiguous in heavily mutated or heavily trimmed sequences, and several equally plausible D alleles are common. Protein input gives V gene assignment only. Chimeric reads from PCR recombination between templates can produce gene combinations that never existed in a cell, so unexpected V-J pairings in amplicon data deserve a second look.