The case for well-conducted experiments to validate statistical protocols for 2D gels: different pre-processing = different lists of significant proteins

Table 9 Comparing t-test results for the three different imputation methods

Normalization 2 Missing replaced by = -17.28 Log transformed (100% of identified spots picked up)	Normalization 2 Missing replaced by Random sample of minimum from 15 gels in each group Log transformed (91% of identified spots picked up)	Normalization 2 Minimum replaced by Random sample of minimum from all 30 gels Log transformed (91% of identified spots picked up)
SSP 1134	SSP1134	SSP 1134
SSP 1509	SSP 1509	SSP 1509
SSP 1733	SSP 1733	SSP 1733
SSP 2309	SSP 2309	SSP 2309
SSP 3219
SSP 3806	SSP 3806	SSP 3806
SSP 4203	SSP 4203	SSP 4203
SSP 4225	SSP 4225	SSP 4225
SSP 4435	SSP 4435	SSP 4435
SSP 4724	SSP 4724	SSP 4724
SSP 5413		SSP 5413
SSP 6205	SSP 6205	SSP 6205
SSP 6304	SSP 6228	SSP 6228
SSP 6314	SSP 6304	SSP 6304
	SSP 6314	SSP 6314
SSP 6452	SSP 6452	SSP 6452
SSP 7223	SSP 7223	SSP 7223
	SSP 7334	SSP 7334
SSP 7750	SSP 7750	SSP 7750

In column 1 missing values were replaced with the lowest intensity value in experiment; in column 2 values to replace missing intensities were randomly chosen from the 15 lowest intensity values within a treatment group; in values to replace missing intensities were randomly chosen from the 30 lowest intensity values without regard to treatment group.

ISSN: 1472-6750