{ "schema": "https://aptatlas.org/data/v4.6/helpMeCombine.schema.json", "name": "helpMeCombine", "title": "Help Me Combine data", "version": "4.6", "last_updated": "2026-11-04", "licence": "CC-BY-4.0", "licence_url": "https://creativecommons.org/licenses/by/4.0/", "citation": "Whelan CD, Smith-Byrne K. Atlas of Proteomic Technologies: an evidence-based framework for selecting and combining commercial proteomics platforms. bioRxiv 2026. https://doi.org/10.64898/2026.09.10.750723", "archive_doi": "10.5281/zenodo.22697191", "third_party": [ { "name": "ChEMBL", "licence": "CC-BY-SA-3.0", "url": "https://www.ebi.ac.uk/chembl/", "relationship": "derived", "note": "The listed fields are derived from ChEMBL (CC BY-SA 3.0): they are counts and percentages computed against a ChEMBL drug-target list. The per-protein list itself is not published here.", "fields": [ "$.universe.drug_targets", "$.universe.drug_targets_on_a_platform", "$.universe.approved_phase4", "$.universe.chembl_fetch_date", "$.coverage_modes[?(@.key=='drug_targets')]", "$.platforms.*.coverage.drug_targets", "$.pairs[*].coverage_pct", "$.pairs[*].coverage_ci_lo", "$.pairs[*].coverage_ci_hi", "$.pairs[*].net_new_targets", "$.pairs[*].net_new_pct", "$.pairs[*].class_balance", "$.anchor_recommendations.*.anchor_solo_pct", "$.anchor_recommendations.*.tiered_partners[*].pair_coverage_pct", "$.anchor_recommendations.*.tiered_partners[*].net_new_targets", "$.anchor_recommendations.*.tiered_partners[*].coverage_gain_pp" ], "fields_derived_in_part": [ "$.pairs[*].composite", "$.pairs[*].pareto_nondominated", "$.pairs[*].top3_pct", "$.pairs[*].mean_regret", "$.pairs[*].win_pct_secondary", "$.pairs[*].tier", "$.anchor_recommendations.*.tiered_partners[*].composite", "$.anchor_recommendations.*.tiered_partners[*].top3_pct", "$.anchor_recommendations.*.tiered_partners[*].mean_regret", "$.anchor_recommendations.*.tiered_partners[*].pareto", "$.anchor_recommendations.*.tiered_partners[*].tier" ] } ], "data": { "universe": { "drug_targets": 1449, "drug_targets_on_a_platform": 1179, "approved_phase4": 929, "proteome_canonical": 20190, "proteome_atlas_union": 14340, "fda_approved_biomarkers": 216, "atlas_fetch_date": "2026-07-26", "chembl_fetch_date": "2026-07-15", "fda_biomarker_source": "Bhowmick P, Roome S, Borchers CH, Goodlett DR, Mohammed Y. J Proteome Res 2021;20(4):2091-2103 (PMC8041396), Table S2, deduplicated to unique UniProt accessions." }, "coverage_modes": [ { "key": "drug_targets", "label": "Drug targets", "denominator": 1449, "default": true }, { "key": "proteome_canonical", "label": "Canonical proteome (SwissProt)", "denominator": 20190 }, { "key": "proteome_atlas_union", "label": "Detectable proteome (Atlas union)", "denominator": 14340 }, { "key": "fda_biomarkers", "label": "FDA-approved biomarkers", "denominator": 216 }, { "key": "ptm_proteoforms", "label": "PTMs / proteoforms", "denominator": null, "note": "Qualitative: mass-spec only. Affinity panels cannot resolve PTMs/proteoforms. Shown as a capability flag, not a %." } ], "quality_dimensions": { "proteome_coverage": 1, "precision": 1, "specificity": 1, "sensitivity": 1, "cost_efficiency": 1, "throughput": 1, "quantification_type": 1, "sample_flexibility": 1, "evidence_depth": 1, "pqtl_accuracy": 1 }, "composite_axes": [ { "key": "coverage", "default_weight": 1, "note": "Coverage denominator set by the coverage-mode selector. No axis is privileged by default; see default_philosophy." }, { "key": "complementarity", "default_weight": 1, "note": "= NET-NEW targets the 2nd platform adds beyond the better single platform (count or % of universe)." }, { "key": "quality", "default_weight": 1, "note": "Default = the five intrinsic measurement properties. Maturity properties separate, opt-in." }, { "key": "cost", "default_weight": 1, "invert": true }, { "key": "class_balance", "default_weight": 0.5 }, { "key": "sample_volume", "default_weight": 0, "note": "Opt-in only." }, { "key": "ms_fit", "default_weight": 0, "note": "Opt-in only. 1 if the pair contains a mass-spectrometry platform (Seer Proteograph XT or Biognosys TrueDiscovery)." } ], "class_balance_rule": "Normalised Shannon entropy of the target classes among the drug targets the pair covers: nine ChEMBL/UniProt classes (target_class_9; GPCRs within membrane receptors, nuclear receptors within transcription factors), Unclassified excluded, divided by ln(9). Derived from the ChEMBL drug-target list. Rounded to 3 decimals.", "quality_axis_dims_used": [ "precision", "specificity", "sensitivity", "quantification_type", "pqtl_accuracy", "throughput", "evidence_depth", "sample_flexibility" ], "dedup_note": "proteome_coverage folded into coverage axis; cost_efficiency into cost axis", "no_feasibility_penalty": "Sample-volume feasibility is NOT auto-penalized. Volume is an optional axis (default weight 0). Infeasible pairs are flagged for the user but not down-scored unless volume is weighted.", "platforms": { "nomic-omni": { "display_name": "Nomic Omni", "coverage_gene_set_size": 1059, "coverage": { "drug_targets": 24, "proteome_canonical": 5.2, "proteome_atlas_union": 7.4, "fda_biomarkers": 41.7 }, "ptm_proteoform_capable": false, "app_scores": { "proteome_coverage": 2, "precision": 4, "specificity": 4, "sensitivity": 3, "cost_efficiency": 5, "throughput": 5, "quantification_type": 4, "sample_flexibility": 2, "evidence_depth": 1, "pqtl_accuracy": 3 }, "sample_volume_uL": 10, "cost_tier_rel": 1 }, "olink-explore-ht": { "display_name": "Olink Explore HT", "coverage_gene_set_size": 5423, "coverage": { "drug_targets": 40.4, "proteome_canonical": 26.9, "proteome_atlas_union": 37.8, "fda_biomarkers": 58.8 }, "ptm_proteoform_capable": false, "app_scores": { "proteome_coverage": 4, "precision": 2, "specificity": 4, "sensitivity": 4, "cost_efficiency": 3, "throughput": 5, "quantification_type": 2, "sample_flexibility": 3, "evidence_depth": 5, "pqtl_accuracy": 3 }, "sample_volume_uL": 2, "cost_tier_rel": 2.5 }, "illumina-protein-prep": { "display_name": "SomaSeq", "coverage_gene_set_size": 9324, "coverage": { "drug_targets": 65.3, "proteome_canonical": 46.2, "proteome_atlas_union": 65, "fda_biomarkers": 90.3 }, "ptm_proteoform_capable": false, "app_scores": { "proteome_coverage": 5, "precision": 5, "specificity": 2, "sensitivity": 4, "cost_efficiency": 3, "throughput": 4, "quantification_type": 2, "sample_flexibility": 3, "evidence_depth": 5, "pqtl_accuracy": 2 }, "sample_volume_uL": 55, "cost_tier_rel": 3 }, "nulisa": { "display_name": "Alamar NULISA", "coverage_gene_set_size": 387, "coverage": { "drug_targets": 10.4, "proteome_canonical": 1.9, "proteome_atlas_union": 2.7, "fda_biomarkers": 14.4 }, "ptm_proteoform_capable": false, "app_scores": { "proteome_coverage": 1, "precision": 5, "specificity": 4, "sensitivity": 5, "cost_efficiency": 4, "throughput": 3, "quantification_type": 2, "sample_flexibility": 3, "evidence_depth": 3, "pqtl_accuracy": 3 }, "sample_volume_uL": 25, "cost_tier_rel": 2 }, "seer-proteograph": { "display_name": "Seer Proteograph XT", "coverage_gene_set_size": 10065, "coverage": { "drug_targets": 57.8, "proteome_canonical": 49.9, "proteome_atlas_union": 70.2, "fda_biomarkers": 88.4 }, "ptm_proteoform_capable": true, "app_scores": { "proteome_coverage": 5, "precision": 3, "specificity": 4, "sensitivity": 3, "cost_efficiency": 3, "throughput": 3, "quantification_type": 3, "sample_flexibility": 3, "evidence_depth": 4, "pqtl_accuracy": 5 }, "sample_volume_uL": 245, "cost_tier_rel": 3 }, "biognosys-truediscovery": { "display_name": "Biognosys TrueDiscovery", "coverage_gene_set_size": 3569, "coverage": { "drug_targets": 33.7, "proteome_canonical": 17.7, "proteome_atlas_union": 24.9, "fda_biomarkers": 73.1 }, "ptm_proteoform_capable": true, "app_scores": { "proteome_coverage": 4, "precision": 3, "specificity": 4, "sensitivity": 3, "cost_efficiency": 2, "throughput": 3, "quantification_type": 3, "sample_flexibility": 5, "evidence_depth": 3, "pqtl_accuracy": 3 }, "sample_volume_uL": 40, "cost_tier_rel": 3 } }, "pairs": [ { "a": "Nomic Omni", "b": "Olink Explore HT", "coverage_pct": 44.3, "coverage_ci_lo": 41.7, "coverage_ci_hi": 47, "coverage_pct_canonical": 27.9, "coverage_pct_atlas_union": 39.2, "coverage_pct_fda": 68.1, "coverage_fda_ci_lo": 62, "coverage_fda_ci_hi": 74.1, "net_new_targets": 56, "net_new_pct": 3.86, "quality_intrinsic": 3.3, "quality_maturity": 3.5, "class_balance": 0.875, "total_cost": 3.5, "joint_volume_uL": 12, "composite": 0.4947, "pareto_nondominated": true, "top3_pct": 32.79, "mean_regret": 0.232, "win_pct_secondary": 14.94, "tier": 2 }, { "a": "Nomic Omni", "b": "SomaSeq", "coverage_pct": 66.5, "coverage_ci_lo": 64.2, "coverage_ci_hi": 68.9, "coverage_pct_canonical": 46.4, "coverage_pct_atlas_union": 65.4, "coverage_pct_fda": 92.1, "coverage_fda_ci_lo": 88.4, "coverage_fda_ci_hi": 95.4, "net_new_targets": 17, "net_new_pct": 1.17, "quality_intrinsic": 3.3, "quality_maturity": 3.33, "class_balance": 0.911, "total_cost": 4, "joint_volume_uL": 65, "composite": 0.5289, "pareto_nondominated": true, "top3_pct": 22.19, "mean_regret": 0.289, "win_pct_secondary": 4.22, "tier": 2 }, { "a": "Nomic Omni", "b": "Alamar NULISA", "coverage_pct": 25.3, "coverage_ci_lo": 23, "coverage_ci_hi": 27.4, "coverage_pct_canonical": 5.7, "coverage_pct_atlas_union": 8, "coverage_pct_fda": 45.4, "coverage_fda_ci_lo": 38.9, "coverage_fda_ci_hi": 52.3, "net_new_targets": 18, "net_new_pct": 1.24, "quality_intrinsic": 3.7, "quality_maturity": 2.83, "class_balance": 0.765, "total_cost": 3, "joint_volume_uL": 35, "composite": 0.4577, "pareto_nondominated": true, "top3_pct": 23.27, "mean_regret": 0.2785, "win_pct_secondary": 8.87, "tier": 2 }, { "a": "Nomic Omni", "b": "Seer Proteograph XT", "coverage_pct": 64.5, "coverage_ci_lo": 62.1, "coverage_ci_hi": 67, "coverage_pct_canonical": 51.2, "coverage_pct_atlas_union": 72, "coverage_pct_fda": 91.7, "coverage_fda_ci_lo": 88, "coverage_fda_ci_hi": 95.4, "net_new_targets": 98, "net_new_pct": 6.76, "quality_intrinsic": 3.6, "quality_maturity": 3, "class_balance": 0.918, "total_cost": 4, "joint_volume_uL": 255, "composite": 0.7178, "pareto_nondominated": true, "top3_pct": 47.51, "mean_regret": 0.16, "win_pct_secondary": 17.97, "tier": 2 }, { "a": "Nomic Omni", "b": "Biognosys TrueDiscovery", "coverage_pct": 43.8, "coverage_ci_lo": 41.3, "coverage_ci_hi": 46.3, "coverage_pct_canonical": 19.9, "coverage_pct_atlas_union": 28.1, "coverage_pct_fda": 79.2, "coverage_fda_ci_lo": 73.6, "coverage_fda_ci_hi": 84.3, "net_new_targets": 146, "net_new_pct": 10.08, "quality_intrinsic": 3.4, "quality_maturity": 3.17, "class_balance": 0.85, "total_cost": 4, "joint_volume_uL": 50, "composite": 0.5784, "pareto_nondominated": true, "top3_pct": 25.43, "mean_regret": 0.2083, "win_pct_secondary": 4.65, "tier": 2 }, { "a": "Olink Explore HT", "b": "SomaSeq", "coverage_pct": 70.5, "coverage_ci_lo": 68, "coverage_ci_hi": 72.8, "coverage_pct_canonical": 54.9, "coverage_pct_atlas_union": 77.4, "coverage_pct_fda": 94.9, "coverage_fda_ci_lo": 91.7, "coverage_fda_ci_hi": 97.7, "net_new_targets": 76, "net_new_pct": 5.24, "quality_intrinsic": 3, "quality_maturity": 4.17, "class_balance": 0.918, "total_cost": 5.5, "joint_volume_uL": 57, "composite": 0.415, "pareto_nondominated": true, "top3_pct": 11.58, "mean_regret": 0.3511, "win_pct_secondary": 3.03, "tier": 3 }, { "a": "Olink Explore HT", "b": "Alamar NULISA", "coverage_pct": 41.3, "coverage_ci_lo": 38.6, "coverage_ci_hi": 43.8, "coverage_pct_canonical": 27.1, "coverage_pct_atlas_union": 38.1, "coverage_pct_fda": 62, "coverage_fda_ci_lo": 55.6, "coverage_fda_ci_hi": 68.1, "net_new_targets": 12, "net_new_pct": 0.83, "quality_intrinsic": 3.4, "quality_maturity": 3.67, "class_balance": 0.867, "total_cost": 4.5, "joint_volume_uL": 27, "composite": 0.3815, "pareto_nondominated": true, "top3_pct": 9.09, "mean_regret": 0.348, "win_pct_secondary": 1.73, "tier": 3 }, { "a": "Olink Explore HT", "b": "Seer Proteograph XT", "coverage_pct": 70, "coverage_ci_lo": 67.8, "coverage_ci_hi": 72.3, "coverage_pct_canonical": 59.6, "coverage_pct_atlas_union": 83.8, "coverage_pct_fda": 94.4, "coverage_fda_ci_lo": 91.2, "coverage_fda_ci_hi": 97.2, "net_new_targets": 177, "net_new_pct": 12.22, "quality_intrinsic": 3.3, "quality_maturity": 3.83, "class_balance": 0.923, "total_cost": 5.5, "joint_volume_uL": 247, "composite": 0.6331, "pareto_nondominated": true, "top3_pct": 33.12, "mean_regret": 0.2187, "win_pct_secondary": 12.88, "tier": 2 }, { "a": "Olink Explore HT", "b": "Biognosys TrueDiscovery", "coverage_pct": 53.7, "coverage_ci_lo": 51.2, "coverage_ci_hi": 56.2, "coverage_pct_canonical": 36.2, "coverage_pct_atlas_union": 51, "coverage_pct_fda": 84.7, "coverage_fda_ci_lo": 79.6, "coverage_fda_ci_hi": 89.4, "net_new_targets": 192, "net_new_pct": 13.25, "quality_intrinsic": 3.1, "quality_maturity": 4, "class_balance": 0.888, "total_cost": 5.5, "joint_volume_uL": 42, "composite": 0.4949, "pareto_nondominated": true, "top3_pct": 20.89, "mean_regret": 0.2467, "win_pct_secondary": 5.95, "tier": 2 }, { "a": "SomaSeq", "b": "Alamar NULISA", "coverage_pct": 65.8, "coverage_ci_lo": 63.4, "coverage_ci_hi": 68.3, "coverage_pct_canonical": 46.3, "coverage_pct_atlas_union": 65.2, "coverage_pct_fda": 91.7, "coverage_fda_ci_lo": 88, "coverage_fda_ci_hi": 94.9, "net_new_targets": 7, "net_new_pct": 0.48, "quality_intrinsic": 3.4, "quality_maturity": 3.5, "class_balance": 0.91, "total_cost": 5, "joint_volume_uL": 80, "composite": 0.471, "pareto_nondominated": true, "top3_pct": 14.5, "mean_regret": 0.3668, "win_pct_secondary": 6.06, "tier": 3 }, { "a": "SomaSeq", "b": "Seer Proteograph XT", "coverage_pct": 78.1, "coverage_ci_lo": 75.9, "coverage_ci_hi": 80.2, "coverage_pct_canonical": 66.3, "coverage_pct_atlas_union": 93.3, "coverage_pct_fda": 98.1, "coverage_fda_ci_lo": 95.8, "coverage_fda_ci_hi": 99.5, "net_new_targets": 185, "net_new_pct": 12.77, "quality_intrinsic": 3.3, "quality_maturity": 3.67, "class_balance": 0.927, "total_cost": 6, "joint_volume_uL": 300, "composite": 0.6424, "pareto_nondominated": true, "top3_pct": 28.25, "mean_regret": 0.2455, "win_pct_secondary": 12.66, "tier": 2 }, { "a": "SomaSeq", "b": "Biognosys TrueDiscovery", "coverage_pct": 69.4, "coverage_ci_lo": 67, "coverage_ci_hi": 71.8, "coverage_pct_canonical": 50.1, "coverage_pct_atlas_union": 70.5, "coverage_pct_fda": 96.3, "coverage_fda_ci_lo": 93.5, "coverage_fda_ci_hi": 98.6, "net_new_targets": 59, "net_new_pct": 4.07, "quality_intrinsic": 3.1, "quality_maturity": 3.83, "class_balance": 0.913, "total_cost": 6, "joint_volume_uL": 95, "composite": 0.3813, "pareto_nondominated": true, "top3_pct": 3.03, "mean_regret": 0.358, "win_pct_secondary": 0.22, "tier": 3 }, { "a": "Alamar NULISA", "b": "Seer Proteograph XT", "coverage_pct": 61.4, "coverage_ci_lo": 58.9, "coverage_ci_hi": 63.9, "coverage_pct_canonical": 50.4, "coverage_pct_atlas_union": 71, "coverage_pct_fda": 90.3, "coverage_fda_ci_lo": 86.1, "coverage_fda_ci_hi": 94, "net_new_targets": 53, "net_new_pct": 3.66, "quality_intrinsic": 3.7, "quality_maturity": 3.17, "class_balance": 0.919, "total_cost": 5, "joint_volume_uL": 270, "composite": 0.6092, "pareto_nondominated": true, "top3_pct": 19.81, "mean_regret": 0.2538, "win_pct_secondary": 6.28, "tier": 3 }, { "a": "Alamar NULISA", "b": "Biognosys TrueDiscovery", "coverage_pct": 39.3, "coverage_ci_lo": 37, "coverage_ci_hi": 41.8, "coverage_pct_canonical": 18.7, "coverage_pct_atlas_union": 26.3, "coverage_pct_fda": 77.8, "coverage_fda_ci_lo": 72.2, "coverage_fda_ci_hi": 83.3, "net_new_targets": 80, "net_new_pct": 5.52, "quality_intrinsic": 3.5, "quality_maturity": 3.33, "class_balance": 0.844, "total_cost": 5, "joint_volume_uL": 65, "composite": 0.4336, "pareto_nondominated": true, "top3_pct": 5.52, "mean_regret": 0.3116, "win_pct_secondary": 0.54, "tier": 3 }, { "a": "Seer Proteograph XT", "b": "Biognosys TrueDiscovery", "coverage_pct": 60.6, "coverage_ci_lo": 58, "coverage_ci_hi": 63.1, "coverage_pct_canonical": 51.1, "coverage_pct_atlas_union": 72, "coverage_pct_fda": 90.7, "coverage_fda_ci_lo": 86.6, "coverage_fda_ci_hi": 94.4, "net_new_targets": 41, "net_new_pct": 2.83, "quality_intrinsic": 3.4, "quality_maturity": 3.5, "class_balance": 0.919, "total_cost": 6, "joint_volume_uL": 285, "composite": 0.4221, "pareto_nondominated": false, "top3_pct": 3.03, "mean_regret": 0.3756, "win_pct_secondary": 0, "tier": 4 } ], "ms_note": "Mass spectrometry is modelled as TWO independent platforms as of engine v2.9 (Seer Proteograph XT and Biognosys TrueDiscovery), each with its own accession-level coverage list from public/proteins.json. Both are independently PTM/proteoform-capable, so ms_fit=1 for any pair containing either.", "fairness_note": "Axis redundancy audit (15 pairs): coverage & class_balance correlate r=+0.93, essentially unchanged from the 10-pair engine. class_balance defaults to weight 0.5 so this redundancy is not double-counted (user-adjustable). No axis is privileged by default (all of coverage/complementarity/quality/cost start at weight 1; see default_philosophy). Under the representative priority sweep (924 scenarios spanning all four numeric coverage denominators), no pair reaches the 80% top-3 threshold required for Tier 1: Nomic Omni + Seer Proteograph XT leads on both win-rate (18.7%) and top-3 rate (48.5%), with SomaSeq + Seer Proteograph XT, Olink Explore HT + Seer Proteograph XT, and Nomic Omni + Biognosys TrueDiscovery clustering close behind as genuine Tier-2 contenders. The ranking is real about being weighting-dependent: which pair 'wins' changes substantially with the user's priorities and chosen coverage denominator (see ranking_sensitivity_note).", "quality_axes": { "intrinsic": { "dims": [ "precision", "specificity", "sensitivity", "quantification_type", "pqtl_accuracy" ], "default_weight": 1, "label": "Intrinsic measurement properties", "note": "How good a single measurement is. THIS IS THE DEFAULT quality axis." }, "maturity": { "dims": [ "throughput", "evidence_depth", "sample_flexibility" ], "default_weight": 0, "label": "Platform maturity / operational", "note": "Adoption & practicality. OFF by default, it reflects how established a platform is, not its measurement properties." } }, "quality_custom_weighting": "Users may override the grouping and set a custom importance weight (0 to 3) on any of the 8 quality sub-dimensions individually. Default = intrinsic dims at 1, maturity dims at 0.", "naming_note": "nELISA is Nomic's technology (= Nomic Omni). NULISA is Alamar's separate platform; different companies, do not conflate. Nomic (nELISA) has the top quantification_type score (4 = absolute pg/mL concentration); NULISA scores 2 (within-assay relative). Seer Proteograph XT and Biognosys TrueDiscovery are both independent mass-spectrometry platforms (see ms_note); they are no longer modelled as a single merged category as of engine v2.9.", "ranking_sensitivity_note": "Under default (equal) weights and the drug-target denominator: Nomic Omni + Seer Proteograph XT ranks highest on the composite (0.7178), followed by SomaSeq + Seer Proteograph XT (0.6424), Olink Explore HT + Seer Proteograph XT (0.6331), and Alamar NULISA + Seer Proteograph XT (0.6092); the gap between #1 and #4 is small relative to the sweep's spread. Across the full priority sweep, max win-share is 18.0% and max top-3 share is 47.5%, both held by Nomic Omni + Seer Proteograph XT; no pair reaches Tier 1 (top-3 >= 80%).", "default_philosophy": "All axes with a nonzero default weight (coverage, complementarity, quality, cost) start EQUAL at weight 1: no single axis is privileged by default. class_balance defaults to 0.5; sample_volume and ms_fit default to 0 and are opt-in only. Users set up to three priorities to raise the axes that matter to them; the tool does not presuppose that coverage (or any other axis) is what a given user cares about most.", "complementarity_note": "Complementarity is measured as NET-NEW targets added, not 1-Jaccard orthogonality, which rewards small panels for a high unique-fraction while contributing few targets in absolute terms. Net-new reflects how many targets the pairing actually lets you measure that a single platform can't. Example: Nomic Omni + Biognosys TrueDiscovery adds 140 net-new drug targets, the largest net-new figure of any Nomic Omni pairing, ahead of Nomic Omni + Seer Proteograph XT (91).", "anchor_recommendations": { "Nomic Omni": { "anchor_solo_pct": 24, "single_suffices": false, "tiered_partners": [ { "partner": "Seer Proteograph XT", "pair_coverage_pct": 64.5, "net_new_targets": 98, "composite": 0.7178, "top3_pct": 47.51, "mean_regret": 0.16, "pareto": true, "coverage_gain_pp": 40.5, "tier": 2 }, { "partner": "Biognosys TrueDiscovery", "pair_coverage_pct": 43.8, "net_new_targets": 146, "composite": 0.5784, "top3_pct": 25.43, "mean_regret": 0.2083, "pareto": true, "coverage_gain_pp": 19.8, "tier": 2 }, { "partner": "SomaSeq", "pair_coverage_pct": 66.5, "net_new_targets": 17, "composite": 0.5289, "top3_pct": 22.19, "mean_regret": 0.289, "pareto": true, "coverage_gain_pp": 42.5, "tier": 2 }, { "partner": "Olink Explore HT", "pair_coverage_pct": 44.3, "net_new_targets": 56, "composite": 0.4947, "top3_pct": 32.79, "mean_regret": 0.232, "pareto": true, "coverage_gain_pp": 20.3, "tier": 2 }, { "partner": "Alamar NULISA", "pair_coverage_pct": 25.3, "net_new_targets": 18, "composite": 0.4577, "top3_pct": 23.27, "mean_regret": 0.2785, "pareto": true, "coverage_gain_pp": 1.3, "tier": 2 } ] }, "Olink Explore HT": { "anchor_solo_pct": 40.4, "single_suffices": false, "tiered_partners": [ { "partner": "Seer Proteograph XT", "pair_coverage_pct": 70, "net_new_targets": 177, "composite": 0.6331, "top3_pct": 33.12, "mean_regret": 0.2187, "pareto": true, "coverage_gain_pp": 29.6, "tier": 2 }, { "partner": "Biognosys TrueDiscovery", "pair_coverage_pct": 53.7, "net_new_targets": 192, "composite": 0.4949, "top3_pct": 20.89, "mean_regret": 0.2467, "pareto": true, "coverage_gain_pp": 13.3, "tier": 2 }, { "partner": "Nomic Omni", "pair_coverage_pct": 44.3, "net_new_targets": 56, "composite": 0.4947, "top3_pct": 32.79, "mean_regret": 0.232, "pareto": true, "coverage_gain_pp": 3.9, "tier": 2 }, { "partner": "SomaSeq", "pair_coverage_pct": 70.5, "net_new_targets": 76, "composite": 0.415, "top3_pct": 11.58, "mean_regret": 0.3511, "pareto": true, "coverage_gain_pp": 30.1, "tier": 3 }, { "partner": "Alamar NULISA", "pair_coverage_pct": 41.3, "net_new_targets": 12, "composite": 0.3815, "top3_pct": 9.09, "mean_regret": 0.348, "pareto": true, "coverage_gain_pp": 0.9, "tier": 3 } ] }, "SomaSeq": { "anchor_solo_pct": 65.3, "single_suffices": true, "tiered_partners": [ { "partner": "Seer Proteograph XT", "pair_coverage_pct": 78.1, "net_new_targets": 185, "composite": 0.6424, "top3_pct": 28.25, "mean_regret": 0.2455, "pareto": true, "coverage_gain_pp": 12.8, "tier": 2 }, { "partner": "Nomic Omni", "pair_coverage_pct": 66.5, "net_new_targets": 17, "composite": 0.5289, "top3_pct": 22.19, "mean_regret": 0.289, "pareto": true, "coverage_gain_pp": 1.2, "tier": 2 }, { "partner": "Alamar NULISA", "pair_coverage_pct": 65.8, "net_new_targets": 7, "composite": 0.471, "top3_pct": 14.5, "mean_regret": 0.3668, "pareto": true, "coverage_gain_pp": 0.5, "tier": 3 }, { "partner": "Olink Explore HT", "pair_coverage_pct": 70.5, "net_new_targets": 76, "composite": 0.415, "top3_pct": 11.58, "mean_regret": 0.3511, "pareto": true, "coverage_gain_pp": 5.2, "tier": 3 }, { "partner": "Biognosys TrueDiscovery", "pair_coverage_pct": 69.4, "net_new_targets": 59, "composite": 0.3813, "top3_pct": 3.03, "mean_regret": 0.358, "pareto": true, "coverage_gain_pp": 4.1, "tier": 3 } ] }, "Alamar NULISA": { "anchor_solo_pct": 10.4, "single_suffices": false, "tiered_partners": [ { "partner": "Seer Proteograph XT", "pair_coverage_pct": 61.4, "net_new_targets": 53, "composite": 0.6092, "top3_pct": 19.81, "mean_regret": 0.2538, "pareto": true, "coverage_gain_pp": 51, "tier": 3 }, { "partner": "SomaSeq", "pair_coverage_pct": 65.8, "net_new_targets": 7, "composite": 0.471, "top3_pct": 14.5, "mean_regret": 0.3668, "pareto": true, "coverage_gain_pp": 55.4, "tier": 3 }, { "partner": "Nomic Omni", "pair_coverage_pct": 25.3, "net_new_targets": 18, "composite": 0.4577, "top3_pct": 23.27, "mean_regret": 0.2785, "pareto": true, "coverage_gain_pp": 14.9, "tier": 2 }, { "partner": "Biognosys TrueDiscovery", "pair_coverage_pct": 39.3, "net_new_targets": 80, "composite": 0.4336, "top3_pct": 5.52, "mean_regret": 0.3116, "pareto": true, "coverage_gain_pp": 28.9, "tier": 3 }, { "partner": "Olink Explore HT", "pair_coverage_pct": 41.3, "net_new_targets": 12, "composite": 0.3815, "top3_pct": 9.09, "mean_regret": 0.348, "pareto": true, "coverage_gain_pp": 30.9, "tier": 3 } ] }, "Seer Proteograph XT": { "anchor_solo_pct": 57.8, "single_suffices": false, "tiered_partners": [ { "partner": "Nomic Omni", "pair_coverage_pct": 64.5, "net_new_targets": 98, "composite": 0.7178, "top3_pct": 47.51, "mean_regret": 0.16, "pareto": true, "coverage_gain_pp": 6.7, "tier": 2 }, { "partner": "SomaSeq", "pair_coverage_pct": 78.1, "net_new_targets": 185, "composite": 0.6424, "top3_pct": 28.25, "mean_regret": 0.2455, "pareto": true, "coverage_gain_pp": 20.3, "tier": 2 }, { "partner": "Olink Explore HT", "pair_coverage_pct": 70, "net_new_targets": 177, "composite": 0.6331, "top3_pct": 33.12, "mean_regret": 0.2187, "pareto": true, "coverage_gain_pp": 12.2, "tier": 2 }, { "partner": "Alamar NULISA", "pair_coverage_pct": 61.4, "net_new_targets": 53, "composite": 0.6092, "top3_pct": 19.81, "mean_regret": 0.2538, "pareto": true, "coverage_gain_pp": 3.6, "tier": 3 }, { "partner": "Biognosys TrueDiscovery", "pair_coverage_pct": 60.6, "net_new_targets": 41, "composite": 0.4221, "top3_pct": 3.03, "mean_regret": 0.3756, "pareto": false, "coverage_gain_pp": 2.8, "tier": 4 } ] }, "Biognosys TrueDiscovery": { "anchor_solo_pct": 33.7, "single_suffices": false, "tiered_partners": [ { "partner": "Nomic Omni", "pair_coverage_pct": 43.8, "net_new_targets": 146, "composite": 0.5784, "top3_pct": 25.43, "mean_regret": 0.2083, "pareto": true, "coverage_gain_pp": 10.1, "tier": 2 }, { "partner": "Olink Explore HT", "pair_coverage_pct": 53.7, "net_new_targets": 192, "composite": 0.4949, "top3_pct": 20.89, "mean_regret": 0.2467, "pareto": true, "coverage_gain_pp": 20, "tier": 2 }, { "partner": "Alamar NULISA", "pair_coverage_pct": 39.3, "net_new_targets": 80, "composite": 0.4336, "top3_pct": 5.52, "mean_regret": 0.3116, "pareto": true, "coverage_gain_pp": 5.6, "tier": 3 }, { "partner": "Seer Proteograph XT", "pair_coverage_pct": 60.6, "net_new_targets": 41, "composite": 0.4221, "top3_pct": 3.03, "mean_regret": 0.3756, "pareto": false, "coverage_gain_pp": 26.9, "tier": 4 }, { "partner": "SomaSeq", "pair_coverage_pct": 69.4, "net_new_targets": 59, "composite": 0.3813, "top3_pct": 3.03, "mean_regret": 0.358, "pareto": true, "coverage_gain_pp": 35.7, "tier": 3 } ] } }, "interaction_modes": { "anchor_mode": { "description": "User selects their FIRST/existing platform. Tool ranks the best SECOND platform to add, showing marginal gain (coverage_gain_pp, net_new_targets) over the anchor alone. This is the primary entry point.", "default_first_platform": "Nomic Omni (consortium's selected anchor), but any platform can be chosen or none." }, "discovery_mode": { "description": "No first platform specified. Tool recommends the best overall pair from the full ranking (currently Nomic Omni + Seer Proteograph XT under default weights).", "trigger": "User picks 'I don't have a platform yet' / leaves anchor unset." }, "single_suffices_flag": "If the anchor alone covers >=60% in the selected mode, flag that a second platform may add limited marginal value (show the marginal gain so the user decides)." }, "fairness_analyses": { "coverage_confidence_intervals": { "method": "2,000-resample bootstrap over the 1,449-target universe (numpy binomial resampling, seed 42), independently regenerated for all 15 pairs; FDA-biomarker intervals use the same method over n=216.", "key_result": "Several pairs remain statistical ties (overlapping 95% CIs); the UI shows CIs and labels overlapping pairs as tied rather than ranking them.", "table": "coverage_bootstrap_ci.csv" }, "pareto_frontier": { "non_dominated": 14, "total": 15, "dominated": [ "Seer Proteograph XT + Biognosys TrueDiscovery" ], "note": "14 of 15 pairs are non-dominated: for nearly every pair there is no strictly-better alternative on all axes, so the 'best' pair genuinely depends on which axis you prioritize. The one dominated pair combines the two mass-spec platforms, whose coverage overlaps heavily and which add limited net-new value relative to an affinity+MS pairing." }, "quality_score_robustness": { "method": "2,000 draws jittering each 1-5 intrinsic quality score by +/-1 (clipped 1-5)", "key_result": "No pair is #1 in anywhere near 100% of perturbations under the new 15-pair field; the ranking is genuinely close among the Tier-2 cluster, which is itself evidence the tool is not silently privileging one pair." }, "robustness_metric": { "primary": [ "top3_pct", "mean_regret" ], "secondary": "win_pct", "why": "Win-rate ('how often is this pair the single best') structurally rewards pairs that PEAK on one axis and penalizes well-rounded pairs. The app reports TOP-3 FREQUENCY and MEAN REGRET as the primary robustness metrics; win-rate is kept only as a secondary 'is there a dominant pick?' signal.", "headline": "Nomic Omni + Seer Proteograph XT leads all three metrics (top-3 48.5%, regret 0.157, win 18.7%) but well short of the 80% top-3 bar for Tier 1, unlike the prior 5-platform engine." } }, "recommendation_output": { "mode": "TIERED, not single-pick", "tiers": { "1": "Leading choice - clear best (top-3 >=80%)", "2": "Strong contenders - genuinely competitive, choice depends on your priority (Pareto-optimal, top-3 >=20%)", "3": "Situational - reasonable only under specific priorities (Pareto-optimal, top-3 <20%)", "4": "Not recommended - dominated (another pair beats them on every axis)" }, "rule": "Always present a TIERED list, never a single winner. In anchor mode, show the anchor's own tiered partner list (anchor_recommendations[anchor].tiered_partners). Flag any T1/T2 gap that is within the coverage CI as a tie.", "consortium_note": "As of engine v2.9 (6 platforms, 15 pairs), no pair reaches Tier 1. For the Nomic Omni anchor, all five partners land in Tier 2 - Nomic Omni's own solo coverage (23.5%) is low enough that any second platform adds substantial value, so the choice among partners depends heavily on which axis (coverage, net-new, or PTM capability) the user prioritizes." }, "engine_version": "v2.11" } }