Glossary

Terms used throughout these pages, in the sense arda uses them.

AIRR Rearrangement

The AIRR Community standard schema for one annotated receptor sequence: gene calls, region coordinates, the junction and its translation, with defined column names and value formats. Every per-sequence TSV arda writes is a valid AIRR Rearrangement file. See Output files and columns.

allele

The most specific level of IMGT gene nomenclature, written with a * suffix: TRBV19*01. Separating a gene’s alleles usually needs substantially more read coverage than naming the gene — see Personalized germline: genotype and call restriction.

amplicon

A targeted library in which nearly every read is receptor-derived and reads span V into J, produced by multiplex V primers or 5’RACE. Sometimes called RepSeq. Handled by arda amplicon.

anchor

The conserved residue that bounds the junction: Cys104 on the V side, Phe or Trp 118 on the J side, in IMGT numbering. arda stores the anchor position and templated residues per allele in cdr3_anchors.tsv and refuses to emit a junction whose V anchor the read never reached.

bulk RNA-seq

A whole-transcriptome library in which 0.02–3 % of reads are receptor-derived and a read lands anywhere in a transcript rather than spanning V into J. Handled by arda rnaseq.

CDR3

Complementarity-determining region 3, in the IMGT definition: the junction excluding both conserved anchors. See junction — the two differ by two residues and conflating them corrupts every downstream coordinate.

chimera

A clonotype whose V and J calls come from loci that cannot rearrange together. arda counts them in Quality control and flags them; it never removes them.

clonotype

A group of reads collapsed onto one rearrangement. The grouping key is set by --clonotype-key: full (locus, V, J, junction) by default, or junction alone. One row of <prefix>.clones.tsv.

consensus_count

AIRR column: the number of distinct fragments supporting a clonotype. Use it when you need molecules. Compare duplicate_count.

duplicate_count

AIRR column: the number of reads encompassing the junction of a clonotype. arda’s expression estimate. Use it when you need abundance. Invariant across every --ec-mode — error correction moves reads onto a parent, it never discards them.

ec-mode

The denoising preset given to arda correct (--ec-mode). Chosen from what the answer will be used for, not from the library type. See Error correction.

fast_fraction

Reported in <prefix>.arda.json: the fraction of reads that hit both a V and a J segment in the two-pass search. It predicts whether the amplicon speed configuration will pay, better than the library’s name does. Roughly .85 on a primer-anchored TCR amplicon, .50 on IGH RepSeq, .05 on bulk.

IMGT

The international ImMunoGeneTics information system: the source of arda’s germline sequences and of the region and numbering conventions it follows.

J+C scaffold

A reference entry consisting of a J germline with the CH1 constant exon spliced onto it, and no V. It gives reads that span the J→C splice — which end in the constant region and have no V to anchor on — somewhere to map, and is what lets arda report a c_call for them. 345 of them ship for human.

junction

The nucleotide (or amino acid) stretch from Cys104 through Phe/Trp118 inclusive of both anchors. This is what AIRR calls junction and what MiXCR labels CDR3. It is two residues longer than IMGT CDR3, which arda writes separately as cdr3.

locus

The receptor chain a sequence belongs to: TRA, TRB, TRG, TRD, IGH, IGK, IGL. arda searches all of them at once and reports the one that won.

mode

One of arda rnaseq, arda amplicon, arda cells. The mode name carries the speed configuration for its library type, because the configurations do not compose. See Usage.

read group

One (R1, R2) FASTQ pair belonging to a sample. A sample delivered as several Illumina lanes is one repertoire made of several read groups; arda maps each and concatenates before clonotype calling. See Samples split across files.

regime

The library class a speed configuration is tuned for — bulk or amplicon. Named once in arda.cluster.regime_flags so the CLI, SLURM, Nextflow and Snakemake cannot disagree.

scaffold

A reference entry made of one V allele and one J allele joined in frame, with a short frame-neutral N spacer where D would sit, carrying IgBLAST-derived FR1–FR4 / CDR1–CDR3 coordinates. Queries are mapped to scaffolds and the markup is projected back onto them. 15,069 V·J scaffolds ship for human.

segment

A single germline allele (V, J or C) in the collapsed 924-target reference that the two-pass search screens against, as opposed to a full V·J scaffold.

SHM

Somatic hypermutation: the AID-driven mutation of rearranged IG loci in germinal centres. arda reports it as v_identity, v_mutations and j_mutations, scoped outside the junction by default. See Somatic hypermutation.

stage

One of the three steps a mode runs: map (Stage 1, per read), correct (Stage 2, global), assemble (Stage 3, global). Each is also a standalone command, which is what makes sharding possible.

tie list

A comma-joined set of gene or allele names that the alignment cannot separate, written into a single v_call / j_call / d_call field. Standard AIRR practice; scoring a tie list as a miss is a scoring artifact rather than a calling error.

UMI

Unique molecular identifier: a random barcode attached to a cDNA molecule before amplification, used to collapse PCR duplicates into one consensus. arda does not collapse UMIs; arda cells consumes the consensus an upstream tool produced.