analytical-method-validation

Run plan_validation.py with the framework, measured attribute (assay, impurity, identity), and technique to get the required characteristics and a protocol skeleton. Q2(R2) supplies almost no numeric acceptance criteria, so you must state them from the specification or analytical target profile before collecting data.

Chromatographic assays use ±15% accuracy/precision tolerances (±20% at LLOQ) with 4 QC levels, while ligand binding assays use ±20% (±25% at LLOQ and ULOQ) with 5 QC levels plus a total-error criterion of 30% (40% at limits). The check_bioanalytical_run.py script requires --modality precisely because these criteria differ.

R-squared rises with range and is insensitive to curvature, so a curved calibration can still show r² above 0.98. The check_response.py script instead uses the lack-of-fit F test against pure error and a runs test on residual signs, which is what ICH Q2(R2) 3.2.2.1 requires.

No, all scripts use only the Python 3.11+ standard library. Statistical distributions (t, chi-square, F) are computed from first principles via the regularized incomplete beta and gamma functions, so results are reproducible in any conforming interpreter.

Use compare_methods.py with paired measurements and a pre-stated equivalence margin; it applies the TOST procedure rather than a t test, since failing to detect a difference is not evidence of equivalence. It also reports Deming and Passing-Bablok regression slopes because ordinary least squares biases the slope when both procedures carry error.

No, the scripts only compute statistics and report findings against stated criteria; they never conclude that a procedure is validated, release a batch, or accept a run. Those decisions remain with the analyst, technical reviewer, and quality unit.