Skip to main content

Table 2 Hook method: algorithm and output characteristics

From: "Hook"-calibration of GeneChip-microarrays: Theory and algorithm

Step Output (chip and probe-set characteristics) Eq.
1) optical background correction using the Affymetrix zone-algorithm (see Ref. [26]). Optical background (O, mean over all zones);
the algorithm uses a scaled value with a scaling factor chosen between 0 and 1
2) Raw hook: Plot of the PM and MM probe intensity data into Δ-vs-Σ coordinates and smoothing over a sliding-window of ~100 probe sets. Classification into N- and S-probes using the breakpoint of the hook. Raw hook curve (9)
3) Parameterization of the positional dependent sensitivity-model separately for the PM and MM in the N and S-ranges and correction of the intensities for probe-specific sensitivities. Sensitivity profiles (optional SN, NN or NNN models) (22), (23); App. C
4) Corrected hook: re-iterate steps (2)–(3) with the corrected intensities to improve the sensitivity correction and the classification of the probes into absent and present ones Corrected hook-curve
Fraction of absent probes (%N), mean N-background level and width of the N-range
5) Fit of the hook-equation to the mix-, S-and sat-ranges of the corrected hook curve and analysis of the probe-level hook coordinates Maximum intensity (Mc), mean non-specific background level (NcPM), dimensions of the hook (αc, βc), PM/MM-affinity gain (sc and nc), parameters of the normal background distribution (μc, σc, ρc) and of the signal distribution (λ), S/N-ratio (R), and occupancy (Θ) and fraction of specific binding (xS) (10), (12), (24) – (26), (29); App. D, App. A
6) Calculation of probe-set related expression estimates (alternatively PMonly or PM-MM) by the joint processing of the intensity data and selected chip characteristics which corrects for the non-specific background, sequence-specific sensitivity and saturation Expression measures (Lset, Sset) (33) – (36)