Some Results for TigrScan

2/12/2004

Speed:

Approx. 12 Kb/sec on a 3.2 GHz Pentium IV  (<3 days to process Human genome).

Sequence Length
  Execution Time (minutes)
100 Kb
 0.15
344 Kb
 0.50
526 Kb
 0.75
840 Kb
 1.22
2.2 Mb
 3.13

Linear correlation coeff: 0.9998

Memory:

Has successfully been used to process sequences over 5.6 Mb in length without any preprocessing.  valgrind-2.0.0 indicates no memory leaks.  Memory usage is linear in sequence length (Linear correlation coeff.: 0.996) :
        
Sequence Length
 Total RAM Usage
300 Kb
 18 Mb
500 Kb
 21 Mb
800 Kb
 24 Mb
2.2 Mb
 36 Mb

Linear correlation coeff: 0.996


Accuracy:

See also: graphic images of toxoplasma predictions vs. curated models.


species
 nuc
sens
%
 nuc
spec
%
 nuc
acc
%
 gt/ag
sens
%
 gt/ag
spec
%
 atg/
tag
sens
 atg/
tag
spec
 exon
sens
%
 exon
spec
%
 exact
genes
%
Toxoplasma
gondii
 85
 97
 94
 72
 78
 48
 47
 62
 66
 19
Aspergillus
fumigatus
 95
 99
 94
 74
 75
 59
 77
 59
 63
 21
Homo sapiens
(very preliminary results)
 58
 78
 96
 39
 72
 33
 31
 34
 55
 4
Arabidopsis
thaliana

96



77
 81
 43
Plasmodium
falciparum
 97
 100
 97
 84
 75
 77
 63
 73
 62
 57
Mus musculus



































































































































































































































































































































































































































































































contact: Bill Majoros