eCAMBer: efficient support for large-scale comparative analysis of multiple bacterial strains

BMC Bioinformatics

Table 6 Overall accuracy statistics for different tools

	PATRIC			Prodigal
Statistic	Input	MA	eCAMBer	Input	GMV	MA	eCAMBer
# of incorrectly removed genes	NA	0	1224	NA	0	0	388
# of incorrectly added genes	NA	1177	792	NA	0	344	331
# of correctly removed genes	NA	0	3993	NA	0	0	1185
# of correctly added genes	NA	410	701	NA	0	210	1447
# of incorrect →correct TIS changes	NA	4812	1591	NA	149	1015	290
# of incorrect →incorrect TIS changes	NA	2223	747	NA	28	1018	113
# of correct →incorrect TIS changes	NA	4279	669	NA	78	3618	170
Precision for gene starts	0.665	0.663	0.699	0.764	0.764	0.734	0.775
Recall for gene starts	0.695	0.702	0.703	0.752	0.753	0.727	0.765
f1 for gene starts	0.680	0.682	0.701	0.758	0.759	0.731	0.770
Precision for gene ends	0.892	0.882	0.920	0.931	0.931	0.928	0.940
Recall for gene ends	0.931	0.935	0.926	0.917	0.917	0.919	0.927
f1 for gene ends	0.911	0.908	0.923	0.924	0.924	0.923	0.934

Overall statistics for accuracy of changes introduced by eCAMBer, Mugsy-Annotator (MA) and the GMV pipeline. The tools were run on the dataset of 20 E. coli with annotations from the PATRIC database (columns 2 to 4) and generated using Prodigal (columns 5 to 8). Correctness of the changes introduced was assessed by comparison with annotations from the Coliscope database. Columns Input correspond to the original annotations. “NA” stands for not applicable. Rows correspond to different statistics of running each tool.

ISSN: 1471-2105