Protein (multi-)location prediction: using location inter-dependencies in a probabilistic framework

Table 4 Multi-location prediction results on the PROSITE-GO version of the dataset, per location, averaged over 25 runs of 5-fold cross-validation, for the combined set of single- and multi-localized proteins

	cyt (3785)	ex (1405)	nuc (2952)	mem (1824)	mi (870)
*Pre- ${Std}_{s_{i}}$ (SVMs)*	0.84 (± 0.01)	0.87 (± 0.02)	0.79 (± 0.02)	0.93 (± 0.01)	0.90 (± 0.03)
*Pre- ${Std}_{s_{i}}$ (Our system)*	0.84 (± 0.01)	0.91 (± 0.02)	0.79 (± 0.03)	0.90 (± 0.01)	0.87 (± 0.03)
*Rec- ${Std}_{s_{i}}$ (SVMs)*	0.85 (± 0.01)	0.64 (± 0.02)	0.72 (± 0.02)	0.79 (± 0.02)	0.62 (± 0.03)
*Rec- ${Std}_{s_{i}}$ (Our system)*	0.86 (± 0.01)	0.65 (± 0.02)	0.74 (± 0.03)	0.80 (± 0.02)	0.66 (± 0.03)
${Pre}_{s_{i}}$ (SVMs)	0.82 (± 0.01)	0.89 (± 0.02)	0.83 (± 0.01)	0.92 (± 0.01)	0.87 (± 0.03)
${Pre}_{s_{i}}$ (Our system)	0.81 (± 0.02)	0.91 (± 0.02)	0.83 (± 0.01)	0.90 (± 0.01)	0.89 (± 0.02)
${Rec}_{s_{i}}$ (SVMs)	0.78 (± 0.01)	0.72 (± 0.02)	0.77 (± 0.01)	0.76 (± 0.01)	0.68 (± 0.02)
${Rec}_{s_{i}}$ (Our system)	0.80 (± 0.01)	0.74 (± 0.02)	0.78 (± 0.02)	0.78 (± 0.01)	0.73 (± 0.02)

Results are shown for the five locations s_i that have the largest number of associated proteins (the number of proteins per location is given in parenthesis): cytoplasm (cyt), extracellular space (ex), nucleus (nuc), membrane (mem), and mitochondrion (mi). The table shows the per-location measures: standard precision (Pre- ${Std}_{s_{i}}$ ), recall (Rec- ${Std}_{s_{i}}$ ), Multilabel-Precision ( ${Pre}_{s_{i}}$ ), and Multilabel-Recall ( ${Rec}_{s_{i}}$ ), obtained from SVMs without using location inter-dependencies and from our system using location inter-dependencies. For each location and measure, the highest of the values obtained from the two methods is shown in boldface. Standard deviations are shown in parentheses.

ISSN: 1748-7188