Metrics in ScreenAssist are assessed at the plate level for every plate in the screen. Plates failing any single criterion should be flagged for review and may require re-screening.
 

Poor Reads (% well scans above fail threshold)

Each well is scanned during a Dianthus uHTS measurement. Individual scans are assigned a quality score based on an ML model, and scans falling below the quality threshold are flagged as “poor reads”. The Poor Reads metric reports the percentage of well scans across a plate that fail this quality threshold.

Poor reads arise from physical issues such as air bubbles and aggregated or precipitated material in the well, or optical artefacts. A high percentage of poor reads indicates systematic sample quality problems or instrument issues, especially when spatially co-located on a plate. Poor reads are excluded from downstream analysis, so a high proportion reduces the effective statistical power of the plate.

Fluorescence CV of Positive Control & Reference Wells

This metric reports the CV of the 650 nm fluorescence intensity signal across all positive control wells and all neutral reference wells on a plate. It measures the reproducibility of the fluorescent target dispense and the consistency of the optical signal.

A CV > 5% indicates that target dispense or stability is inconsistent. Since the Spectral Shift ratio (650 nm / 670 nm) is the primary readout, high fluorescence variability at 650 nm will degrade ratio precision and reduce assay sensitivity.

Neutral RSDE (Robust Standard Deviation Equivalent)

In the context of acceptance criteria, the Neutral RSDE threshold ensures the neutral reference population has sufficient tightness for reliable hit identification. 

• RSDE > 0.02 (Fail): The neutral reference is too variable for confident Z score-based hit calling.

• RSDE ≤ 0.008 (Pass): The neutral reference population is tight and reproducible.

🔗Neutral RSDE (Robust Standard Deviation Equivalent) in Statistical Foundations for the full definition. 

Robust Z Prime & Z Prime

Both metrics are assessed per plate using the industry-standard thresholds (≥ 0.5 excellent).

Plates in the marginal zone (0.4–0.5) should be reviewed carefully: hits identified on these plates carry greater uncertainty and should be prioritized for orthogonal confirmation assays.

When Robust Z’ and classical Z’ diverge, outlier removal or investigation should be conducted before making final quality decisions.

🔗Z Prime (Z’) and Robust Z Prime (Robust Z’) in Statistical Foundations for full definitions.

|SSMD| (Strictly Standardized Mean Difference)

An |SSMD| ≥ 6 corresponds to a very large effect size with a very low false positive/negative rate.

SSMD values below 6 indicate that the positive control and neutral reference populations overlap substantially, and hit calling confidence will be reduced. SSMD is particularly informative when Z’ values appear borderline, as it provides an independent probabilistic assessment of assay quality.

🔗 Strictly Standardized Mean Difference (SSMD) in Statistical Foundations for the full definition. 

Titration Fitted Curve S/N

The signal-to-noise ratio (S/N) of the fitted titration curve is derived from the dose-response curve fit performed during assay development. It quantifies the ratio of the total signal window (fitted response amplitude) to the measurement noise. 🔗 Signal-to-noise ratio

A high S/N confirms that the assay has a robust signal window over the detection noise and that the fitted titration curve is reliable. An S/N ≥ 12 (excellent) provides high confidence in the affinity measurement and in the suitability of the assay for compound screening.

Titration Response Amplitude (Δ Ratio: bound vs. unbound)

The Titration Response Amplitude is the absolute difference in the Spectral Shift ratio between the fully bound state (at saturating ligand concentration) and the unbound state (no ligand) of the fitted titration curve. It reports the size of the assay signal window available for distinguishing binders from non-binders. 🔗 Response Quality

A Δ Ratio ≥ 0.018 is required for a valid assay. This threshold ensures the response amplitude is large enough to detect partial binders at single-dose screening concentration. An amplitude below this threshold means the signal change upon binding is too small relative to noise, increasing the likelihood of false negatives and reducing dynamic range.

Δ 650 nm Fluorescence: Positive Control vs. Neutral Reference (Single Point)

This metric monitors the percentage change in 650 nm raw fluorescence intensity between the mean of positive control wells and the mean of neutral reference wells in single-point screening mode.

The 650 nm channel reports total dye fluorescence intensity, which should remain stable between conditions if dye loading is consistent. A large Δ fluorescence indicates a systematic difference in dye loading or sample conditions between the two control populations, which can occur due to:

• Aggregation or precipitation of the protein in one condition.

• Autofluorescence of the positive control ligand at 650 nm.

• Quenching of the dye fluorescence by the positive control compound (inner filter effect).

A Δ fluorescence > 40% will confound the ratio-based Spectral Shift measurement and renders the plate unreliable for hit identification. Values in the marginal zone (≤ 40%, > 20%) warrant investigation before proceeding.

Δ Fluorescence of Titration: 650 nm counts at bound vs. unbound states

Analogous to the single-point metric Δ 650 nm Fluorescence Positive Control vs. Neutral Reference (Single Point), but evaluated across the full titration curve. The 650 nm fluorescence should ideally remain constant across the titration concentration series; the signal change should be in the ratio (relative shifts between 650 nm and 670 nm emission), not in absolute 650 nm intensity. 🔗 Response Quality

A large Δ fluorescence across the titration (≥ 40%) indicates that the compound or titration conditions are directly affecting dye fluorescence, for example, through:

• Competition with the dye binding site.

• Protein aggregation at high compound concentrations.

• Inner filter effects at high compound concentrations.

This makes the assay window unreliable and the Spectral Shift ratio uninformative for hit identification.