Chapter · August 2026 · IBM Heron r2

Pharma Atlas
multi-logic drug decisions

Drug research lives in conflict: efficacy against safety, exposure against toxicity, speed against selectivity. QPC keeps those logics co-resident in one quantum pass and returns a verified ranking on IBM hardware.

Pharma is the proving ground for QPC

Simultaneous, contradictory objectives are the work of drug research. A lead must enter the body, stay long enough, miss the heart channel, spare the liver, and still hit the target. QPC holds those logics co-resident in one quantum job. Transjunctions couple them while the computation runs. The readout is a joint ranking, with public IBM job IDs.

QPC is built for that full complexity in a single pass. Several contextures occupy the same quantum job at once. That is simultaneous parallel computing at the level of the architecture, on IBM Heron r2, 156 qubits.

Below are four verified real-world results. Each desk froze a public panel, locked an expectation before submit, ran one coupled job, and scored the claim on raw counts. The polypharmacy safety desk sits with them on Real-World Desks.

August 2026 · Discovery · Architecture WOW

ADMET LEAD WOW · PASS

Seventeen conflicting ADMET logics kept co-resident in one 156-qubit job on ibm_marrakesh. Classical weighted-sum and sequential filters disagree on the frozen 9×17 table. QPC closed a pre-registered architecture scorecard the same day: raw junction gap 0.185, Aer retention 0.749, separable 0.184. Geometry lock held.

Claim PASS ibm_marrakesh · 156Q · 8192 shots raw junction gap 0.185 Aer retention 0.749
ADMET chapter page → WOW article → Lead desk report →

August 2026 · Discovery · Pre-registered claim

KINASE SELECTIVITY 9×17 · PASS

Real-world selectivity choice on IBM Heron r2 ibm_kingston: locked expectation that imatinib outranks dasatinib when intended hits, collateral hits, and safety kinases co-reside, with raw junction gap at least 0.05. Verified on raw QPC readout against equal-weight potency on the same Davis 2011 panel. Pair claim plus architecture gap. Full panel ranks are reported on the case page.

Claim PASS ibm_kingston · Heron r2 · 156Q da632keaa69c739lgpag 0.610 > 0.547 · gap 0.197
Kinase selectivity report → How it was run →

August 2026 · Oncology · Pre-registered claim

MELANOMA 1L BRAF+ · PASS

Real-world first-line choice for BRAF V600+ advanced melanoma on IBM Heron r2 ibm_kingston: locked expectation that nivo+ipi outranks nivo+rela under co-resident efficacy / safety / burden. Verified on raw QPC readout against equal-weight classical. Pair claim on the co-resident ranking. Full panel ranks are on the case page.

Claim PASS ibm_kingston · Heron r2 da1gvd6g52gs73clkp20 0.569 > 0.394
Melanoma 1L BRAF+ report → How it was run →

August 2026 · Pharma · Pre-registered claim

PDT THERAPY F2 · PASS

Real-world photodynamic therapy selection on IBM Kingston: locked expectation that talaporfin outranks padeliporfin under co-resident efficacy / toxicity / selectivity. Verified on raw QPC readout against equal-weight classical. Therapy panel on Kingston, public job ID.

Claim PASS ibm_kingston da0bkvv2sl0c73bmtf2g 0.546 > 0.493
PDT Therapy F2 report → How it was run →