← TAA Surface-Target Prioritization
Discovery / Target Biology · Target biologist / antibody & cell-therapy discovery scientist (surface-antigen selection)

TAA Surface-Target Prioritization — agent trajectory for expert review

gpt-5.2-chat-latest · 2026-06-04T05:34:52.620567+00:00 · 4 tool calls · 19.1s · 5763 tokens
run id: taa_prioritization_20260603_223432_01_evaluate-dll3-as-a-tumor-associated-antigen-for

Query

Evaluate DLL3 as a tumor-associated antigen for small cell lung cancer (SCLC).

Trajectory

Agent · turn 1
⚙ calls list_universe()
↳ list_universe result
show data
{
  "genes_available": [
    "CD19",
    "CEACAM5",
    "CLDN18",
    "DLL3",
    "EGFR",
    "EPCAM",
    "ERBB2",
    "FOLR1",
    "GAPDH",
    "GPC3",
    "MS4A1",
    "MSLN",
    "MUC1",
    "PTPRC",
    "TACSTD2",
    "TP53"
  ],
  "tumor_cohorts": [
    "BRCA",
    "STAD",
    "LUAD",
    "LUSC",
    "COAD",
    "OV",
    "PAAD",
    "LIHC",
    "HNSC",
    "GBM",
    "SCLC",
    "DLBC",
    "KIRC",
    "PRAD"
  ],
  "normal_tissues": [
    "Heart - Left Ventricle",
    "Brain - Cortex",
    "Lung",
    "Liver",
    "Kidney - Cortex",
    "Colon - Transverse",
    "Stomach",
    "Pancreas",
    "Nerve - Tibial",
    "Whole Blood",
    "Spleen",
    "Skin - Sun Exposed",
    "Esophagus - Mucosa",
    "Thyroid",
    "Testis",
    "Ovary",
    "Breast - Mammary Tissue",
    "Prostate"
  ],
  "vital_tissues": [
    "Brain - Cerebellum",
    "Brain - Cortex",
    "Colon - Sigmoid",
    "Colon - Transverse",
    "Heart - Atrial Appendage",
    "Heart - Left Ventricle",
    "Kidney - Cortex",
    "Liver",
    "Lung",
    "Nerve - Tibial",
    "Pancreas",
    "Spleen",
    "Stomach",
    "Whole Blood"
  ],
  "modality_context": "Targets are evaluated for cell-surface therapeutics (CAR-T, ADC, bispecific / T-cell engager). These can only engage an EXTRACELLULAR epitope, so a credible cell-surface domain is a hard requirement. Expression in vital/dose-limiting normal tissues drives on-target/off-tumor toxicity.",
  "expression_units": "TPM (transcripts per million)"
}
Agent · turn 2
⚙ calls get_localization(gene=DLL3)
⚙ calls get_tumor_expression(gene=DLL3)
⚙ calls get_normal_expression(gene=DLL3)
↳ get_localization result
show data
{
  "gene": "DLL3",
  "accession": "Q9NYJ7",
  "protein_name": "Delta-like protein 3",
  "localizations": [
    "Golgi apparatus",
    "Cell membrane"
  ],
  "topology": "Single-pass type I membrane protein",
  "transmembrane_count": 1,
  "note": "Predominantly Golgi/intracellular; low cell-surface density limits ADC but works for T-cell engager (tarlatamab)."
}
↳ get_tumor_expression result
show data
{
  "gene": "DLL3",
  "tumor_tpm": {
    "BRCA": 1.0,
    "STAD": 1.0,
    "LUAD": 2.0,
    "LUSC": 2.0,
    "COAD": 1.0,
    "OV": 1.0,
    "PAAD": 3.0,
    "LIHC": 0.5,
    "HNSC": 1.0,
    "GBM": 8.0,
    "SCLC": 60.0,
    "DLBC": 0.5,
    "KIRC": 0.5,
    "PRAD": 3.0
  }
}
↳ get_normal_expression result
show data
{
  "gene": "DLL3",
  "normal_tpm": {
    "Heart - Left Ventricle": 0.1,
    "Brain - Cortex": 2.0,
    "Lung": 0.3,
    "Liver": 0.1,
    "Kidney - Cortex": 0.1,
    "Colon - Transverse": 0.1,
    "Stomach": 0.1,
    "Pancreas": 1.0,
    "Nerve - Tibial": 0.5,
    "Whole Blood": 0.1,
    "Spleen": 0.1,
    "Skin - Sun Exposed": 0.1,
    "Esophagus - Mucosa": 0.1,
    "Thyroid": 0.2,
    "Testis": 1.0,
    "Ovary": 0.2,
    "Breast - Mammary Tissue": 0.1,
    "Prostate": 0.2
  }
}
Agent · turn 3

Target: DLL3 (Delta-like protein 3)

Indication: Small Cell Lung Cancer (SCLC)


1. Surface Eligibility (hard requirement)

Localization (UniProt evidence):
- Subcellular localizations: Golgi apparatus, Cell membrane
- Topology: Single-pass type I membrane protein
- Transmembrane domains: 1
- Note: Predominantly Golgi/intracellular; low cell-surface density limits ADC but works for T-cell engager (tarlatamab).

Interpretation

DLL3 is a single-pass type I membrane protein, which structurally implies:
- An extracellular N-terminal domain
- A single transmembrane helix
- A cytoplasmic C-terminal tail

Thus, DLL3 is formally surface-eligible, because it possesses an extracellular domain.

However, UniProt notes that it is predominantly Golgi/intracellular, with low cell-surface density. This is a critical nuance:
- CAR-T / TCE (high sensitivity, serial killing) → more tolerant of low density
- ADCs → may be more limited by low antigen density

Conclusion: Surface eligible, but surface density may be limiting depending on modality.


2. On-Tumor Signal (TCGA SCLC)

Tumor median TPM:

Comparator cohorts:
- LUAD: 2.0
- LUSC: 2.0
- GBM: 8.0
- Most others: 0.5–3.0

Interpretation

DLL3 expression in SCLC (60 TPM) is:
- ~30× higher than LUAD/LUSC (2 TPM)
- ~60× higher than many epithelial tumors (1 TPM)
- Substantially higher than GBM (8 TPM)

This is a very strong, lineage-restricted tumor signal within the TCGA dataset.

Excellent on-tumor signal for SCLC


3. Tumor vs Normal Specificity (Dominant Criterion)

Normal tissue median TPM (selected key tissues):

Tumor (SCLC): 60.0 TPM

Tumor-to-normal ratios:

Interpretation

Given:
- Brain is a dose-limiting organ
- DLL3 is mostly intracellular in normal tissues
- Absolute TPM is modest

Overall specificity profile is favorable.

Excellent tumor-vs-normal differential


4. Safety Considerations

Vital tissues (selected):

Tissue TPM
Brain – Cortex 2.0
Lung 0.3
Heart – LV 0.1
Liver 0.1
Kidney 0.1
Pancreas 1.0
Whole Blood 0.1

Key observations:
- No high expression in heart, liver, kidney, lung, or blood.
- Brain expression is present (2 TPM), but still low relative to tumor (60 TPM).
- Hematologic tissues are minimal → low risk of marrow aplasia.

Given low normal expression and predominantly intracellular localization in non-tumor contexts, on-target/off-tumor toxicity risk appears manageable, particularly for controlled modalities (e.g., bispecifics).


FINAL ASSESSMENT

Target: DLL3
Indication: Small Cell Lung Cancer (SCLC)

Surface Eligibility

Eligible for cell-surface therapeutics.
Evidence: Single-pass type I membrane protein with cell membrane localization and an extracellular domain. UniProt notes predominant Golgi localization and low surface density, which may limit ADC performance but remains compatible with CAR-T or T-cell engager approaches.

On-Tumor Signal

Specificity (Tumor vs Normal — dominant criterion)

This represents a strong and selective tumor expression pattern.

Safety

Tier

Tier 1 – Strong

Rationale and Recommendation

DLL3 demonstrates high SCLC-restricted expression (60 TPM) with low expression across vital normal tissues (≤2 TPM), yielding an excellent tumor-to-normal differential. It is structurally a surface protein with an extracellular domain, satisfying the hard requirement for CAR-T, ADC, and T-cell engager modalities, though its predominantly intracellular localization suggests that modality choice matters (T-cell engagers/CAR-T likely superior to ADC).

Overall, DLL3 is a high-priority tumor-associated antigen for SCLC, with strong specificity and acceptable safety margins. It qualifies as a Tier 1 target for cell-surface immunotherapeutic strategies in SCLC.

Final output

Target: DLL3 (Delta-like protein 3)

Indication: Small Cell Lung Cancer (SCLC)


1. Surface Eligibility (hard requirement)

Localization (UniProt evidence):
- Subcellular localizations: Golgi apparatus, Cell membrane
- Topology: Single-pass type I membrane protein
- Transmembrane domains: 1
- Note: Predominantly Golgi/intracellular; low cell-surface density limits ADC but works for T-cell engager (tarlatamab).

Interpretation

DLL3 is a single-pass type I membrane protein, which structurally implies:
- An extracellular N-terminal domain
- A single transmembrane helix
- A cytoplasmic C-terminal tail

Thus, DLL3 is formally surface-eligible, because it possesses an extracellular domain.

However, UniProt notes that it is predominantly Golgi/intracellular, with low cell-surface density. This is a critical nuance:
- CAR-T / TCE (high sensitivity, serial killing) → more tolerant of low density
- ADCs → may be more limited by low antigen density

Conclusion: Surface eligible, but surface density may be limiting depending on modality.


2. On-Tumor Signal (TCGA SCLC)

Tumor median TPM:

Comparator cohorts:
- LUAD: 2.0
- LUSC: 2.0
- GBM: 8.0
- Most others: 0.5–3.0

Interpretation

DLL3 expression in SCLC (60 TPM) is:
- ~30× higher than LUAD/LUSC (2 TPM)
- ~60× higher than many epithelial tumors (1 TPM)
- Substantially higher than GBM (8 TPM)

This is a very strong, lineage-restricted tumor signal within the TCGA dataset.

Excellent on-tumor signal for SCLC


3. Tumor vs Normal Specificity (Dominant Criterion)

Normal tissue median TPM (selected key tissues):

Tumor (SCLC): 60.0 TPM

Tumor-to-normal ratios:

Interpretation

Given:
- Brain is a dose-limiting organ
- DLL3 is mostly intracellular in normal tissues
- Absolute TPM is modest

Overall specificity profile is favorable.

Excellent tumor-vs-normal differential


4. Safety Considerations

Vital tissues (selected):

Tissue TPM
Brain – Cortex 2.0
Lung 0.3
Heart – LV 0.1
Liver 0.1
Kidney 0.1
Pancreas 1.0
Whole Blood 0.1

Key observations:
- No high expression in heart, liver, kidney, lung, or blood.
- Brain expression is present (2 TPM), but still low relative to tumor (60 TPM).
- Hematologic tissues are minimal → low risk of marrow aplasia.

Given low normal expression and predominantly intracellular localization in non-tumor contexts, on-target/off-tumor toxicity risk appears manageable, particularly for controlled modalities (e.g., bispecifics).


FINAL ASSESSMENT

Target: DLL3
Indication: Small Cell Lung Cancer (SCLC)

Surface Eligibility

Eligible for cell-surface therapeutics.
Evidence: Single-pass type I membrane protein with cell membrane localization and an extracellular domain. UniProt notes predominant Golgi localization and low surface density, which may limit ADC performance but remains compatible with CAR-T or T-cell engager approaches.

On-Tumor Signal

Specificity (Tumor vs Normal — dominant criterion)

This represents a strong and selective tumor expression pattern.

Safety

Tier

Tier 1 – Strong

Rationale and Recommendation

DLL3 demonstrates high SCLC-restricted expression (60 TPM) with low expression across vital normal tissues (≤2 TPM), yielding an excellent tumor-to-normal differential. It is structurally a surface protein with an extracellular domain, satisfying the hard requirement for CAR-T, ADC, and T-cell engager modalities, though its predominantly intracellular localization suggests that modality choice matters (T-cell engagers/CAR-T likely superior to ADC).

Overall, DLL3 is a high-priority tumor-associated antigen for SCLC, with strong specificity and acceptable safety margins. It qualifies as a Tier 1 target for cell-surface immunotherapeutic strategies in SCLC.

Reference answer key (deterministic) — open after grading

Deterministic TAA scoring model (config.py / scoring.py): base = 100·(0.25·tumor_expression + 0.45·tumor/normal_ratio + 0.30·surface_quality), then ×safety multiplier (0.15-1.0, 70% vital-tissue expression + 30% breadth of normal expression). HARD GATE: no surface ECD -> SURFACE_INELIGIBLE, score capped at 15.0. Tiers: >=60 Tier 1, >=40 Tier 2, >=25 Tier 3, else Tier 4. A strong agent reaches the same gating and ordering from the raw tool data — especially rejecting intracellular antigens (TP53, GAPDH) regardless of tumor expression, and reading vital-tissue expression as a dose-limiting liability.

GeneCohortScore (0-100)TierTumor / mean-normal TPM (ratio)Flags
DLL3SCLC82.4Tier 1 - strong60.0 / 0.36 = 44.3xLOW_SURFACE_DENSITY

Expert grading

Surface-accessibility gating 1 (poor) 5 (excellent)
5: Treats a credible cell-surface extracellular domain as a hard gate: correctly deprioritizes intracellular antigens (TP53, GAPDH) as surface-ineligible for CAR-T/ADC/bispecific regardless of tumor expression, and notes they are only addressable via TCR-T/pMHC.
1: Recommends or fails to disqualify an intracellular antigen for a surface modality, or ignores localization/topology entirely and scores on tumor expression alone.
Tumor/normal specificity 1 (poor) 5 (excellent)
5: Judges the antigen on tumor expression RELATIVE to normal-tissue expression (tumor/normal ratio), not raw tumor TPM; recognizes that a high tumor TPM with high normal expression is a poor therapeutic window.
1: Equates high tumor expression with a good target, ignores normal-tissue baseline, or never computes/considers the tumor-vs-normal ratio.
Normal-tissue / vital-organ safety 1 (poor) 5 (excellent)
5: Flags expression in vital, dose-limiting tissues (heart, brain, lung, liver, etc.) as an on-target/off-tumor toxicity liability and weighs breadth of normal expression; treats vital-tissue expression as a dose-limiting gate, not a minor deduction.
1: Misses or downplays vital-tissue expression, treats safety as an afterthought, or never checks the GTEx normal profile against the vital-tissue list.
Integrated prioritization & tiering 1 (poor) 5 (excellent)
5: Synthesizes surface gate + specificity + safety into a clear, well-ordered prioritization (score/tier or ranked recommendation) that matches the evidence, with calibrated confidence and the right caveats; in comparisons, orders candidates correctly.
1: No clear prioritization, a ranking contradicted by its own evidence, or over/under-confidence inconsistent with the data.
Evidence faithfulness 1 (poor) 5 (excellent)
5: Every number and claim (TPM values, topology, ratios, flags) traces to the tool outputs; no fabricated expression values, variants, or programs.
1: Hallucinates TPM numbers, localization, or clinical programs, or contradicts the returned tool data.
Overall verdict
Your name / ID