Paramus — SAR Series Analysis

Overview

Given analogs + activity, find the scaffold, group by R-substitution, correlate changes with activity, and surface activity cliffs. The value is the SAR narrative, not a single descriptor call.

Guidance vs execution: alignments, matched pairs, and descriptors come from tested Paramus tools. Do not eyeball SAR from model memory — ground it in the analysis.

When to Use

• A congeneric series with measured activity and "what drives potency?"
• Identifying matched molecular pairs and activity cliffs Do not use for: de novo generation (hand off to a generative workflow).

Workflow

Analog set + activity values
  ↓ 1. Parse & standardize          → Paramus cheminformatics tool
  ↓ 2. Scaffold / core detection    → common core + R-group decomposition
  ↓ 3. Matched molecular pairs      → Paramus MMP tool
  ↓ 4. Correlate change ↔ activity  → per-substituent effect + activity cliffs
  ↓ 5. Report                       → SAR table, cliffs, driving substituents + provenance

Procedure

1. Discover tools via search/get_schema ("scaffold decomposition", "matched molecular pairs").
2. Standardize structures; confirm they share a core (report outliers).
3. Decompose into core + R-groups; enumerate matched pairs.
4. Correlate each substitution with the activity delta; flag activity cliffs (large ΔpIC50, small Δstructure).
5. Report the SAR table, cliffs, and the substituents that drive/kill activity, with data provenance.

Domain Judgment

• Correlation is not causation — present trends as hypotheses to test, not mechanisms.
• Activity cliffs are the highest-information points; highlight them explicitly.
• Respect assay variability: don't over-interpret deltas within measurement noise.

Fallbacks

• MMP tool unavailable → deliver R-group + trend analysis, clearly scoped.
• Endpoint unreachable → stop and report; no local SAR heuristics.