Skip to main content

Table 5 Shift Score Summary

From: Improving pairwise sequence alignment accuracy using near-optimal protein sequence alignments

 

Alignments

Better (count)

Worse (count)

Better (avg. (stdev))

Worse (avg. (stdev))

High Similarity

model trim vs. optimal

3

1

0.011 (0.005)

-0.009 (0.000)

 

robust trim vs. optimal

3

1

0.010 (0.006)

-0.009 (0.000)

 

frequency trim vs. optimal

3

1

0.010 (0.006)

-0.009 (0.000)

 

local trim vs. optimal

13

0

0.022 (0.017)

0

 

local trim + model trim vs. local

0

10

0

-0.008 (0.003)

 

model alignment vs. optimal

8

3

0.025 (0.035)

-0.008 (0.002)

 

robust alignment vs. optimal

3

12

0.019 (0.014)

-0.140 (0.130)

 

frequency alignment vs. optimal

6

3

0.028 (0.038)

-0.008 (0.002)

Medium Similarity

model trim vs. optimal

7

3

0.036 (0.025)

-0.013 (0.006)

 

robust trim vs. optimal

9

4

0.032 (0.025)

-0.010 (0.005)

 

frequency trim vs. optimal

7

4

0.037 (0.025)

-0.011 (0.006)

 

local trim vs. optimal

14

2

0.088 (0.076)

-0.014 (0.001)

 

local trim + model trim vs. local

6

10

0.019 (0.010)

-0.017 (0.013)

 

model alignment vs. optimal

10

2

0.034 (0.030)

-0.016 (0.013)

 

robust alignment vs. optimal

2

19

0.095 (0.062)

-0.200 (0.150)

 

frequency alignment vs. optimal

9

2

0.035 (0.030)

-0.016 (0.013)

Low Similarity

model trim vs. optimal

7

3

0.032 (0.022)

-0.022 (0.015)

 

robust trim vs. optimal

11

3

0.027 (0.020)

-0.022 (0.016)

 

frequency trim vs. optimal

7

3

0.032 (0.022)

-0.022 (0.015)

 

local trim vs. optimal

18

3

0.100 (0.096)

-0.240 (0.390)

 

local trim + model trim vs. local

5

8

0.023 (0.019)

-0.013 (0.007)

 

model alignment vs. optimal

7

2

0.025 (0.014)

-0.026 (0.019)

 

robust alignment vs. optimal

3

16

0.088 (0.066)

-0.220 (0.180)

 

frequency alignment vs. optimal

6

3

0.023 (0.014)

-0.024 (0.014)

  1. Numbers shown are the number of sequence pairs, out of 22 in each similarity group, whose shift score with respect to the Dali alignment are improved or worsened by at least 0.005, along with the average and standard deviation of the magnitude of the change in shift score. In the case of local alignments or alignments limited to the local alignment boundaries, the Dali alignment is also limited to the local alignment boundaries. Model trim is the optimal alignment, with edges that fall below 50% probability according to the logistic regression model removed. Robust trim is the same, except edges less than a normalized robustness score of 0.5 are removed. Frequency trim is for edges with a frequency less than 0.5 removed. Local trim is the optimal (semi-global) alignment, with all edges outside the SSEARCH boundaries removed. Model alignment is an alignment produced using the log-odds score produced by the full logistic regression model in place of the substitution matrix scores. Robust and frequency alignment is the same, except the logistic regression model only used robustness or frequency as a variable.