Skip to main content

Table 2 Experimental design

From: Comparison of methods for calculating conditional expectations of sufficient statistics for continuous time Markov chains

GY

      

Experiment

 

2

  

3

 

k

t k

μt k

s ( μ tk )

t k

μt k

s ( μ tk )

1

0.0017

0.0045

5

0.1

0.2668

8

2

0.0032

0.0085

5

0.2

0.5337

9

3

0.0046

0.0124

5

0.3

0.8005

11

4

0.0061

0.0163

5

0.4

1.0674

12

5

0.0076

0.0202

5

0.5

1.3342

13

6

0.0090

0.0241

5

0.6

1.6010

14

7

0.0105

0.0281

6

0.7

1.8679

15

8

0.0120

0.0320

6

0.8

2.1347

15

9

0.0135

0.0359

6

0.9

2.4015

16

10

0.0150

0.0398

6

1.0

2.6684

17

GTR

      

Experiment

 

5

  

6

 

k

t k

μt k

s(μt k )

t k

μt k

s(μt k )

1

0.1760

0.2668

8

0.1

0.1516

7

2

0.3520

0.5337

9

0.6

0.9098

11

3

0.5280

0.8005

11

1.1

1.6680

14

4

0.7039

1.0674

12

1.6

2.4262

16

5

0.8798

1.3342

13

2.1

3.1844

18

6

1.0558

1.6010

14

2.6

3.9426

20

7

1.2318

1.8679

15

3.1

4.7008

22

8

1.4077

2.1347

15

3.6

5.4590

24

9

1.5837

2.4015

16

4.1

6.2172

26

10

1.7597

2.6684

17

4.6

6.9754

27

UNR

      

Experiment

 

7

  

8

 

k

t k

μt k

s(μ tk )

t k

μt k

s(μ tk )

1

0.0379

0.1516

7

0.1

0.4

9

2

0.2275

0.9098

11

0.6

2.4

16

3

0.4170

1.6680

14

1.1

4.4

21

4

0.6066

2.4262

16

1.6

6.4

26

5

0.7961

3.1844

18

2.1

8.4

30

6

0.9857

3.9426

20

2.6

10.4

34

7

1.1752

4.7008

22

3.1

12.4

38

8

1.3648

5.4590

24

3.6

14.4

42

9

1.5543

6.2172

26

4.1

16.4

45

10

1.7439

6.9754

27

4.6

18.4

49

  1. The table shows the time points t k , μ tk and the approximation of the Poission tail s(μt k ). For experiment 2, t k spanned the interval that contains the 10 longest branch lengths from the phylogeny of the 16 HIV pol sequences. In experiment 3 we started at 0.1 and ended at 1. We wished to design experiment 5 such that the corresponding s(μt k ) was the same as s(μt k ) from experiment 3. This allowed us to illustrate the relative performance of the methods when running on different sizes of the rate matrix. Experiment 6 was done on time points starting from 0.1 and ending at 4.6. As before, we wished to design experiment 7 such s(μt k ) corresponded to experiment 6. Experiment 8 used the same t k as experiment 6.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us