Summary of first round reannotation results.

The column ‘svm reannotation’ includes the proteins identified by SVM classification reannotation method; the column ‘random svm’ includes the proteins randomly drawn; the column ‘clt reannotation’ includes the proteins identified by hierarchical clustering for reannotation; the column ‘random clt’ includes the proteins randomly drawn. The row ‘AM right’ indicates the proteins whose automated classified or clustered annotations were right, while the previous human annotations were wrong; the row ‘partially right’ indicates the proteins whose automated annotations were partially right, in that reannotation added the predicted annotation to the previous one; the row ‘both right’ indicates the proteins whose automated annotations were the same as previous human annotations, and where reannotation did not change it; the row ‘AM wrong’ indicates the proteins whose automated annotations were wrong, while the previous human annotations were right; the row ‘both wrong’ indicates the proteins whose automated annotations and previous human annotations were wrong, and a new assignment was made during reannotation. ‘Negative’ indicates those proteins that were reannotated as ‘non-specific location’ and designated for removal from the next release of the dataset. They could correspond to bad antibodies.