pytigger

Pure-Python port of the R/CRAN package tigger — Tools for Immunoglobulin Genotype Elucidation via Rep-seq.

tigger is part of the Immcantation framework (Kleinstein Lab, Yale). It discovers novel immunoglobulin V alleles from adaptive immune receptor repertoire sequencing data (AIRR-Seq / Rep-Seq), infers a subject's V genotype, and corrects V-allele calls accordingly.

pytigger is a faithful, dependency-light re-implementation of tigger 1.1.3 in pure Python (numpy / scipy / pandas / matplotlib, no rpy2). Numerical parity with the R package is the design priority: on the bundled example data the novel-allele table, genotype membership and reassigned calls match R exactly, and the Bayesian likelihood numbers agree to a relative difference below 1e-13.

Installation

pip install pytigger

From source:

git clone https://github.com/omicverse/py-tigger
cd py-tigger
pip install -e .

The TIgGER trifecta

import pytigger as tg

# Bundled example data (the datasets shipped with R tigger)
data = tg.load_airrdb()                 # 17,559 AIRR-seq sequences
germ = tg.load_sample_germline_ighv()   # 344 IGHV germline alleles

# 1. Discover novel V alleles
novel = tg.find_novel_alleles(data, germ)
tg.select_novel(novel)[["germline_call", "polymorphism_call", "note"]]
#   germline_call  polymorphism_call               note
#   IGHV1-8*02     IGHV1-8*02_G234T   Novel allele found!

# 2. Infer the subject's V genotype
geno = tg.infer_genotype(data, germline_db=germ, novel=novel,
                         find_unmutated=True)
# ... or the full Bayesian model
geno_b = tg.infer_genotype_bayesian(data, germline_db=germ, novel=novel)

# 3. Correct the V-call assignments
gdb = tg.genotype_fasta(geno, germ, novel)
out = tg.reassign_alleles(data, gdb)        # adds 'v_call_genotyped'

# Evidence table for the inferred novel alleles
ev = tg.generate_evidence(out, novel, geno, gdb, germ)

Visualisation

fig1 = tg.plot_novel(data, tg.select_novel(novel).iloc[[0]])  # evidence plot
fig2 = tg.plot_genotype(tg.load_sample_genotype())            # genotype grid

API

The trifecta + core

function	R equivalent
find_novel_alleles	findNovelAlleles
select_novel	selectNovel
infer_genotype	inferGenotype
infer_genotype_bayesian	inferGenotypeBayesian
reassign_alleles	reassignAlleles
genotype_fasta	genotypeFasta
generate_evidence	generateEvidence
get_popular_mutation_count	getPopularMutationCount

Sequence / IO utilities

read_ig_fasta, write_fasta, clean_seqs, update_allele_names, sort_alleles, get_mutated_positions, get_mut_count, find_unmutated_calls, insert_polymorphisms, super_substring.

Segment parsing (alakazam port)

get_gene, get_allele, get_family, translate_dna.

Plotting

plot_novel, plot_genotype.

Built-in data

load_airrdb, load_sample_germline_ighv, load_sample_novel, load_sample_genotype.

R-parity

pytigger is validated against tigger 1.1.3 (installed alongside alakazam 1.4.3). Running the trifecta on the bundled AIRRDb / SampleGermlineIGHV example data:

• find_novel_alleles — all 30 output columns match R for all 12 rows; the single novel allele IGHV1-8*02_G234T is detected identically.
• infer_genotype — gene/allele/count membership and the "Cannot distinguish" notes match R exactly.
• infer_genotype_bayesian — discrete genotype matches R; the kh, kd, kt, kq, k_diff log-likelihoods agree to rel-diff < 1e-13.
• reassign_alleles — the v_call_genotyped column matches R on 100% of 17,559 sequences.
• generate_evidence — the evidence metrics match R exactly.

See tests/test_r_parity.py (skipped automatically when no R is available).

Citation

If you use pytigger, please cite the original tigger papers:

Gadala-Maria D, Yaari G, Uduman M, Kleinstein SH (2015). Automated analysis of high-throughput B-cell sequencing data reveals a high frequency of novel immunoglobulin V gene segment alleles. PNAS 112(8):E862-70.

Gadala-Maria D, Gidoni M, Marquez S, et al. (2019). Identification of Subject-Specific Immunoglobulin Alleles from Expressed Repertoire Sequencing Data. Front. Immunol. 10:129.

License

AGPL-3, retained from the original tigger package. See LICENSE.