Skip to main content

Table 4 Real data application

From: Impact of adaptive filtering on power and false discovery rate in RNA-seq experiments

 

No filter

Basic

Mean-based

Max-based

Zero-based

Jaccard

Bottomly

1443

1324 (35)

1488 (24)

1489 (25)

1417 (15)

1371 (34)

Sultan

-

2801 (83)

3511 (11)

3445 (15)

3511 14)

2864 (42)

Airway

0

1029 (48)

1554 (60)

1576 (60)

1235 (26)

1612 (57)

Airway 2

-

102 (52)

275 (80)

347 (80)

120 (16)

197 (54)

Mouse

9151

8772 (21)

9173 (6)

9540 (18)

9192 (5)

8593 (28)

Kidney

13076

13075 (3)

13282 (5)

13299 (4)

11784 (19)

13072 (2)

Kidney 2

5777

6357 (5)

6357 (2)

6355 (3)

6357 (4)

6355 (3)

Yuen

      

gill vs. mantle

7932

7932 (6)

9912 (38)

10340 (48)

8619 (25)

7932 (0)

gill vs. foot

7534

7534 (8)

9537 (38)

9685 (32)

8194 (27)

7543 (0)

gill vs. vmass

6093

6093 (4)

8079 (44)

8948 (55)

7023 (23)

6093 (0)

mantle vs. foot

5291

5291 (12)

5802 (41)

5911 (36)

5470 (13)

5291 (0)

mantle vs. vmass

2468

2468 (6)

3655 (47)

4007 (68)

3034 (23)

2468 (0)

foot vs. vmass

3605

3605 (7)

5054 (35)

5602 (66)

4178 (27)

3605 (0)

  1. Maximum number of rejections for each filtering method and the corresponding observed proportion of filtered genes in parentheses (for the basic filter based on all genes, for other filters on the non-zero genes) for several data sets, multiplicity adjustment with lfdrs, \(\alpha =0.05\) Filtering is performed at the end (order (a)). The adaptive filter with the highest number of rejections is highlighted in bold