Table 4

Crumble results for 90 kb of genomic DNA from seven species.

Time

Log-likelihood1


Pecan2

11.3

-0.354

Crumble w/Pecan

60%

7.42

-0.355

30%

4.67

-0.357

15%

5.42

-0.357


FSA3

38.3

-0.374

Crumble w/FSA

60%

20.4

-0.375

30%

12.2

-0.375

15%

9.68

-0.376


MUSCLE4

_a

_a

Crumble w/MUSCLE

60%

_a

_a

30%

153.

-0.363

15%

59.2

-0.367


1 The log-likelihood of the alignment as calculated by phyloFit, in millions of nats.

2 Pecan was run with default parameters.

3 FSA was run with the --exonerate, --anchored, --softmasked, and --fast flags.

4 MUSCLE was run with default parameters.

5 This problem was unable to be aligned due to running out of memory.

The run-time and log-likelihood score of Crumble alignments. Each underlying alignment method (Pecan, FSA, MUSCLE) was tested on the dataset. Crumble was then used to break the problem into sub-problems that were approximately 60%, 30%, and 15% of the length of the original problem. While MUSCLE was unable to align this problem directly, using Crumble we were able to apply it to this problem.

Roskin et al. BMC Bioinformatics 2011 12:144   doi:10.1186/1471-2105-12-144

Open Data