Automatic categorization of diverse experimental information in the bioscience literature

BMC Bioinformatics

Table 4 Evaluation results of nine FlyBase data types With low occurrence using the testing sets

Data type	Recall	Filter term (%)
Initial characterization of a gene	0.97 ± 0.05	18.0 ± 1.3
Use of expression marker	0.95 ± 0.06	22.5 ± 2.3
Transfection of DNA/RNA	0.94 ± 0.04	7.6 ± 1.6
New phenotype (characterization)	0.93 ± 0.05	19.9 ± 2.1
Renaming of a gene	0.91 ± 0.10	10.9 ± 2.6
New cis-regulatory elements	0.88 ± 0.05	8.1 ± 2.2
Gene model modification	0.88 ± 0.08	17.1 ± 3.5
Genome feature sequence mapping	0.87 ± 0.09	10.9 ± 2.6
Merge of gene reports	0.86 ± 0.06	13.7 ± 5.3

The SVM analysis was done using training/testing sets specified in Additional File 7, Table S6 and Methods.

ISSN: 1471-2105