A reusable framework for biotech partnering analysis.

Coverage and conventions

A curated biopharma transaction dataset covering licenses, collaborations, non-exclusive licenses, asset purchases and joint ventures. Records are not all pure licensing transactions; a separate transaction-type field is not available.

The collection's stated coverage focus is Upfront ≥ US$50m. Upfront may include cash and equity, which remain separate components. Combined “upfront and milestones” amounts must not be interpreted as upfront. Currency and payment definitions must be verified before judging eligibility; the application imports the main transaction sheet as supplied, without silently excluding records on an unverified conversion.

597 transactions; announcement dates 2018-01-24 to 2026-09-03. Data review version: 2026-09-30. Content version: 97d4a27c274c (SHA-256 prefix). Only Page 1 transactions are analyzed. Internal review fields and other worksheets are not shown or exported.

Date means the public announcement of the agreement, not closing or publication of a later article. Purchaser / licensee is displayed at the recorded parent-company level. Company names are taken from the curated workbook; the application does not guess new parent mappings. Original supplied values and previous versions remain available to the operator for review.

Matching and missing data

Only selected, positive-weight criteria enter the calculation. Therapeutic area and modality use an application-only multi-label mapping. All matching labels are retained, and each transaction appears only once per search. Selected TA restricts the cohort to records carrying that label (unless its weight is zero). Modality and stage contribute to the existing similarity score; missing values remain unknown. Source TA text is mapped to broad categories; hepatology maps to Gastrointestinal, and unclassified non-missing text maps to Other. Target-only or route-only descriptions do not establish modality. Indication uses literal, case-insensitive phrase matching. Stage retains the five standardized simplified source categories; result rows show the original historical phase. The recorded stage is not independently verified as the exact stage on the signing date.

Similarity = matched weight / weight of known comparable fields. Not Disclosed and blank fields are unknown, not mismatches. Weighted coverage = known-field weight / all selected-field weight. At least 50% coverage and the chosen similarity threshold are required. A 100% threshold means all known, weighted fields match; it does not require all selected fields to be known.

High: similarity ≥ 75% and coverage ≥ 75%. Medium: similarity ≥ 50% and coverage ≥ 50%. Other qualifying records are Low. Results rank by label, then similarity, coverage and newest announcement. Similarity is not deal quality or a probability of signing. Default weights (adjustable from 0 to 100):

• Therapeutic area: 30
• Development stage: 25
• Modality: 20
• Indication: 15
• Licensor country: 3
• Licensee country: 2

Deal terms and benchmarks

Source amounts are original-currency millions (owner confirmed). Non-USD conversion awaits exchange-rate direction confirmation; only USD records enter dollar benchmarks.

All transactions meeting the current threshold and coverage rule form the benchmark cohort, independent of the displayed page (5, 10 or 20 per page); all matching transactions are accessible through pagination. Each metric reports its own valid n. Numeric non-negative values, including zero, are included; Not Disclosed values are never treated as zero. Text and unavailable USD conversions are excluded. Linear interpolation produces P25, median and P75. Samples below five are flagged; percentiles below three observations are withheld. Upfront / total excludes a zero total.

Upfront and total potential use the recorded Upfront and All-in value fields. Upfront may comprise cash plus equity. These components are displayed separately as recorded, without recomputing or silently overriding the upfront field. A combined upfront-and-milestones figure is not upfront. Inclusion of upfront in total, royalty treatment, exclusivity and single-asset versus package scope are not separately confirmed. Existing notes and cash, equity and milestone figures are shown in the row detail without invented interpretations. Amounts are descriptive, not valuation estimates.

Historical partners and supporting views

Partners are purchasers / licensees in the complete qualifying cohort. The top 20 rank by mean similarity, mean coverage, then relevant transaction count and name. Parent-company names follow the supplied purchaser / licensee field; aliases are not speculatively consolidated. Each card shows recorded activity and up to five supporting transactions drawn only from the current page of comps. Historical deal activity does not imply current interest, strategic fit or willingness to transact.

The optional pathway view prefers at least three precedents with similarity and coverage each ≥ 75%; otherwise it uses the qualifying cohort. It describes stage, rights and counterparty patterns, with sample-size labels rather than statistical confidence. Market patterns describe the cohort, not the market population. There is no NPV, M&A assessment or transaction-structure recommendation.

Sources, exports and privacy

Prefer SEC filings and company announcements when verifying deals. Existing HTTP(S) links in allowed transaction fields are displayed when present; source links are never invented. Missing links are explicitly identified. The current workbook has no dedicated source-link column. Internal notes, image references and other worksheets are excluded from both result details and CSV exports. Source links found only in excluded material are not published.

CSV exports contain the displayed comps, full-cohort aggregate benchmarks, or selected partner cards, plus search criteria, weights, scope, data version and export date. They are generated on the server from the authorized current search. No full-workbook download is provided. A shortlist lasts for the current result page and resets on reload or a new search. Your current search is held in the eight-hour server session for export authorization and included in your CSV; pilot evidence logs retain counts, not search text. Sign-out deletes the session; subsequent sign-ins clean expired sessions. Handle exports as private evaluation material.

Outputs are informational only and are not investment, legal, or transaction advice. Independently verify source facts and assumptions before relying on any result.