Adaptive Immunity

IN_PROGRESS BIOLOGY_DOMAINFLAGSHIP

Species: human

Adaptive Immunity

Bottom line: this project covers the adaptive immune system in human:
how antigen is processed and shown on MHC, how T and B cell receptors are built
by V(D)J recombination and signal, how co-stimulation and checkpoints set the
response, how T and B cells differentiate, and how antibodies are diversified
and act. We divided it into 12 areas, each grounded in GO biological process
terms, with 110 human anchor genes. The T cell receptor area is finished: all
19 of its genes have COMPLETE reviews, in two batches. The first covered the
receptor trunk (CD3D, CD3E, CD3G, LCK, ZAP70, LAT, LCP2, PLCG1, NFATC1;
1,103 existing annotations plus 10 new). The second covered the remaining
pathway members (CD4, CD8B, PTPRC, GRAP2, ITK and the calcineurin subunits
PPP3CA, PPP3CB, PPP3R1), with CD8A brought into line. The curated T cell
receptor signaling module now contains all 19 proteins in five steps, from
antigen recognition to NFAT-driven transcription. Across the project, 38 of the
110 genes have a review (30 COMPLETE). The main calls were to replace generic
protein binding rows with the activity each paper shows, to record
coreceptors and the CD3 chains as contributors to the receptor complex rather
than its subunits or bearers of its activity, and to keep knockout-phenotype
processes non-core. MHC class I and II presentation, V(D)J recombination and T
cell cytotoxicity still have no reviews; the MHC class I pathway is next.

The counts come from coverage.md, which
ADAPTIVE_IMMUNITY/scripts/coverage.py generates from the scope file
ADAPTIVE_IMMUNITY/areas.yaml, the gene reviews, the modules and
gocams/index.tsv. Re-run the script after new reviews land; do not edit the
counts by hand.

Scope

In scope: processes that depend on somatically rearranged antigen receptors
(GO:0002460, under GO:0002250 adaptive immune response), and the machinery
that builds, triggers, regulates and executes them. Out of scope: innate
pattern recognition, inflammasomes, interferon induction and complement,
except where a gene is also an adaptive anchor (for example the Fc receptors
that carry out antibody effector function). Innate-immune work already has its
own projects: cGAS-STING,
NLRP3 inflammasome and
C. elegans surveillance immunity.

Areas

Reviewed / anchor genes per area, from coverage.md.

Area GO grounding Modules Reviewed
MHC class I antigen processing and presentation GO:0019885 none 0/8
MHC class II antigen processing and presentation GO:0019886 none 0/7
V(D)J recombination GO:0033151 none 0/7
T cell receptor signaling GO:0050852 t_cell_receptor_signaling 19/19
T cell co-stimulation and inhibitory checkpoints GO:0031295 none 3/9
IL-2 and common gamma-chain cytokine signaling GO:0038110 jak_stat_signaling 5/11
Helper and regulatory T cell differentiation GO:0046632 none 5/9
T cell mediated cytotoxicity GO:0001913 none 0/7
B cell receptor signaling GO:0050853 b_cell_receptor_signaling 5/12
B cell differentiation and germinal center GO:0030183, GO:0002467 none 1/10
Class switch recombination and somatic hypermutation GO:0045190, GO:0016446 none 1/6
Antibody transport and Fc receptor effector signaling GO:0006959 Fc-gamma, Fc-epsilon 0/7

A gene can anchor more than one area (BCL6 and IRF4 each appear
twice), so the per-area gene counts add up to 112, not 110.

Existing resources

T cell receptor trunk: findings

Gene Annotations Accept Non-core Over-annotated Modify Remove Undecided New
human/CD3D 62 45 14 2 0 1 0 1
human/CD3E 106 64 7 11 13 9 2 0
human/CD3G 111 88 12 1 1 9 0 0
human/LCK 234 54 95 9 8 64 4 0
human/ZAP70 98 52 17 1 12 16 0 0
human/LAT 87 55 8 3 14 7 0 3
human/LCP2 109 39 1 1 37 30 1 2
human/PLCG1 230 135 16 1 54 21 3 2
human/NFATC1 66 45 7 1 7 6 0 2
Total 1,103 577 177 30 146 163 10 10

"Annotations" counts existing GOA rows; "New" rows are additional.

Second batch: completing the pathway

Curation rules for adaptive-immune genes

  1. Separate the activity from the pathway. Give each protein its own
    molecular function (peptide antigen binding for MHC, ABC-type peptide
    antigen transporter activity for TAP, protein tyrosine kinase activity for
    LCK/ZAP70/SYK, a scaffold activity for LAT/BLNK, endonuclease activity for
    RAG1 and DCLRE1C, DNA cytosine deamination for AICDA) and use the pathway terms as process. Replace generic
    protein binding IPI rows with an informative term or remove them; they
    were 144 of the 174 removals in AUTOIMMUNE.
  2. Pleiotropic DNA repair genes are non-core here. PRKDC, LIG4, NHEJ1, UNG,
    MSH2, MSH6, POLH and REV1 act in V(D)J recombination, class switching or
    hypermutation, but their core function is general DNA repair. Keep the
    adaptive process terms as non-core; the lymphocyte-specific enzymes (RAG1,
    RAG2, DNTT, AICDA) are the core members.
  3. Knockout phenotypes record necessity, not participation. Mouse
    knockouts put terms such as T cell differentiation, B cell proliferation
    or immunoglobulin production on almost any gene whose loss blocks
    lymphocyte development. Ask which protein performs the step before accepting
    or proposing such a term, per CLAUDE.md.
  4. Ligands, antigens and substrates do not perform the step. A peptide is
    not involved_in antigen processing, and a cytokine is not involved_in
    the receptor signaling it triggers unless it does part of the work. Run the
    comparator check from CLAUDE.md before proposing such a term with NEW.
  5. Check propagation carefully. MHC, KIR, Fc receptor and immunoglobulin
    families expand and diverge by lineage, and many adaptive genes have no
    ortholog outside jawed vertebrates. Review ISO and IBA rows against the
    ortholog relation and node placement (IBA review,
    ISO failure taxonomy).
  6. Viral immune evasion is annotated on the viral protein. The TAP and LAT
    inhibitors above get symbiont-mediated suppression terms; the host protein
    gets no term for being targeted.

Plan