|
Comparison between MAS4 and MOID. |
||
| MAS4 |
MOID |
|
|
|
||
| Mismatch used |
Yes |
No |
| Sensitivity to probe set size |
Super-Olympic filtering is turned off when remaining probe number is less than 8. |
Confidence intervals gradually increase as probe set size drops. |
| Relative error (Spiking tests) |
30% |
28% |
| Minimum of probes required (Spiking tests) |
10 |
10 |
| No-change test results (True fold change = 1.0) |
80% of fold change values in [0.80, 1.34] |
80% of fold change values in [0.92,1.22] |
| Normalization |
Constant based on average intensity. Normalize once. |
Allow both constant and curve normalization. Iterative scheme. |
| Problematic genes (Bright mismatches) |
May underestimate expression level |
May overestimate expression level |
| Expression Level |
Average difference (may be negative). No confidence interval evaluations |
Expression value (always positive) Lower bound and upper bound (with probe set size taken into account. |
| Present Call |
A, P, M (discrete, unstable). Rules were derived from 50 μm and 24 μm feature arrays, may need update for 20 μm) |
Continuous LogP |
Zhou and Abagyan BMC Bioinformatics 2002 3:3 doi:10.1186/1471-2105-3-3 |
||