Table 4.

Examples: The Selective Expression Algorithm Applied to Synthetic, yet Realistic, Data

Example Baseline adjusted λ[ii] gap[iii] τ[iv] log10(sp)[v] d [vi] Comments
1ano0.550.67−6.260.33reference example
1byes1.000.550.67−6.270.33same as 1a; λ has no effect[vii]
2ano0.280.68−6.260.24 d different from 1a due to  gap only
2byes0.990.280.68−6.280.24 d different from 1a due to  gap; λ has no effect[vii]
3ano0.280.68−6.260.24 d different from 1a due to  gap only
3byes0.870.280.58−4.900.00 d different from 1a due to  λ−adjusted log10(sp) > −5,  hence d = 0[viii]

[i] The example identification (1, 2, or 3) corresponds to Fig. 9; whether a baseline compression adjustment was omitted (a) or used (b) in the discordancy computation (equation 6 or 11). The intensities and source quality weights are from Table 3.

[vi] Equation { label needed for disp-formula[@id='E15'] }. Equation { label needed for disp-formula[@id='E9'] } sigmoidal parameters areb = 10 and c = 0.8. The parameter values in the decision function d (equations 13–15) are α = β = γ = 1.5, δ = 0.3, g thresh = 0.25, log10(sp)thresh = −5, and log10(sp) = −20.

[vii] λ has no effect, as the baseline (i.e., ∼0.3, Table 3) is distant from the maximum allowed intensity (i.e., 1).

[viii] λ has non-negligible effect, as the baseline is near (i.e., 0.67, Table 3) the maximum intensity.