Python package for high-precision inference of lineages in antibody repertoires (HILARy).
This package implements the methods described in Combining mutation and recombination statistics to infer clonal families in antibody repertoires, including:
-
A priori estimation of prevalence, the fraction of pairs in the dataset linking sequences belonging to the same clonal family (module
apriori.py). -
Fast CDR3-based clustering with fixed precision/sensitivity (class
CDR3Clusteringin moduleinference.py). -
Full method relying on information encoded in the CDR3 as well as phylogenetic signal encoded outside the CDR3 (class
HILARyin moduleinference.py). -
Evaluation of inference results (module
aposteriori.py)
The dependencies can be installed with
pip install sonnia (see soNNia)
pip install atriegc (see ATrieGC)
The input are aligned sequences in AIRR-compatible format, a tab-separated file with the following columns
| Name | Example |
|---|---|
| sequence_id | 1 |
| v_call | IGHV1-2*01 |
| j_call | IGHJ1*01 |
| junction | TGTCATGCGATTAACAGCGCGTGG |
| v_sequence_alignment | TCTGACGACACGGCCGTATATTACTGT |
| j_sequence_alignment | TGGGGCCGGGGGACC |
| v_germline_alignment | TCTGACGACACGGCCGTGTATTACTGT |
| j_germline_alignment | TGGGGCCAGGGCACC |
See inference.ipynb for example pipeline.