High type I error and misrepresentations in search for transgenerational epigenetic inheritance: response to Guerrero-Bosagna

Table 1 Power calculations based on primary data in [1]. Unadjusted power and total sample size calculations were done in R v3.2.2 (via the pwr.t.test function in the pwr package [7]. Effect sizes were Cohen’s d. The difference in means was the corresponding log2 fold change. SD was estimated as the pooled standard deviation for a given comparison. These power calculations did not assume false discovery rate (FDR) adjustments because we were concerned that it would be overly conservative and remove potential true positives (tests were not necessarily independent). Further, the inclusion of the false positives that the FDR corrections would have removed should actually increase our odds of identifying any transgenerational effects (whether they were true or false positives), but even under these more relaxed conditions, none could be established. FDR corrected calculations are also displayed in the last two columns. These were done in R via the ssize.twoSamp function from the ssize.fdr package [8]. A true positive ratio of 0.05 was assumed for the FDR calculations

Comparison number	Sample size (n treated + n control)	Pooled AVG STDV	Power for log(2) change	Power to detect 50 % change	Effect size for 50 % difference power calculation	Number of samples to detect 1.5-fold change with power = 0.9	Number of samples to detect 1.5-fold change with power = 0.8	Number of samples to detect 1.2-fold change with power = 0.9	Number of samples to detect 1.2-fold change with power = 0.8	Number of samples to have 5 % FDR in detecting 1.2-fold change	Number of samples to have 5 % FDR in detecting 1.05-fold change
1	6	0.16	1.00	0.92	3.75	6	6	18	14	32	370
2	6	0.16	1.00	0.91	3.64	6	6	18	14	34	392
3	6	0.15	1.00	0.94	3.96	6	6	16	14	30	332
4	6	0.16	1.00	0.90	3.57	8	6	20	16	34	408
5	6	0.15	1.00	0.93	3.80	6	6	18	14	32	362
6	6	0.16	1.00	0.90	3.56	8	6	20	16	34	410
13	4	0.15	0.89	0.55	3.93	6	6	16	14	30	338
14	4	0.15	0.88	0.53	3.78	6	6	18	14	32	364
17	6	0.15	1.00	0.93	3.83	6	6	18	14	30	354
18	6	0.16	1.00	0.92	3.71	6	6	18	14	32	378
19	6	0.14	1.00	0.97	4.26	6	6	14	12	26	288
20	6	0.15	1.00	0.94	3.90	6	6	16	14	30	342
21	6	0.15	1.00	0.95	4.02	6	6	16	12	28	322
22	6	0.15	1.00	0.93	3.84	6	6	18	14	30	352
29	4	0.15	0.89	0.55	3.91	6	6	16	14	30	340
30	4	0.16	0.85	0.49	3.57	8	6	20	16	34	406

ISSN: 1474-760X