Labreador

Labreador automates the analysis of microplate bioassay data through client-side computational engines that perform nonlinear regression, statistical group comparison, parallel line analysis, and ecotoxicological risk assessment directly within the user's browser.

Key Features:

• Dose-Response Curve Fitting: Advanced regression models including 4-parameter (4PL) and 5-parameter (5PL) logistic equations, coupled with an automated multi-start optimization strategy and Levenberg-Marquardt algorithm, with weighted regression for heteroscedasticity correction and interactive outlier management.
• ELISA Standard Curve Analysis: Fits ELISA standard curves (4PL, 5PL, linear, quadratic, log-log) with back-calculated recovery and quality checks.
• IC50/EC50 Estimation: Estimates IC50/EC50 from dose-response data with Z' factor calculation.
• Parallel Line Analysis (PLA): Relative potency determination following Ph. Eur. 5.3 and USP <1032>/<1034>, including joint 4PL/5PL parallel fit, F-test for parallelism, Fieller's theorem and bootstrap 95% confidence intervals, quality gate, and multi-run combination with chi-squared homogeneity test and weighted geometric mean.
• Statistical Group Comparison: Automated decision trees utilizing Welch's ANOVA and non-parametric alternatives (Kruskal-Wallis) to prevent Type I error inflation, with t-test, ANOVA, post-hoc and non-parametric tests for kinetic plate-reader data reduction.
• Ecotoxicology Risk Assessment: Specialized indices for pollution, bioaccumulation, and risk assessment including EDI, THQ, HI, CR, CF, PLI, Igeo, BSAF, BAF, and Fulton's K.
• Quality Control System: Three-level quality control with automatic blocking of interpolation when calibration curve quality is insufficient, strict NaN propagation for missing reference data, and enforced interpolation limits to prevent false-safe classifications.

Scientific Applications:

• ELISA Quantification: Analysis of enzyme-linked immunosorbent assay data with standard curve fitting and sample concentration determination.
• Dose-Response Studies: Pharmacological and toxicological assessment of compound potency through IC50/EC50 determination with confidence intervals.
• Relative Potency Assays: Bioequivalence and comparability studies following pharmacopeial guidelines for parallel line analysis.
• Kinetic Microplate Analysis: Time-course data reduction from plate-reader experiments with group statistics and hypothesis testing.
• Environmental Risk Assessment: Computation of ecotoxicological indices for pollution monitoring, bioaccumulation evaluation, and human health risk characterization.

Methodology:

Computations are performed client-side using the Levenberg-Marquardt optimization algorithm for nonlinear regression, automated model selection from five regression models, weighted regression for heteroscedasticity correction, and automated decision trees for statistical testing. Parallel line analysis employs joint 4PL/5PL parallel fit with F-test for parallelism, Fieller's theorem and bootstrap methods for confidence intervals, and chi-squared homogeneity testing for multi-run combination.