Table 3. Open in new tab Performance of... | Oxford Academic

Table 3.

Open in new tab

Performance of RNALoc-LM and its variant models on independent test sets for three types of RNAs.^a

RNA type	Variant model	ACC	Macro F1	Macro precision	Macro recall
lncRNA	Without RNA-FM (one-hot)	0.641	0.418	0.376	0.510
	Without RNA-FM (word2vec)	0.579	0.418	0.385	0.494
	Without CNN	0.660	0.579	0.576	0.608
	Without Bi-LSTM	0.674	0.571	0.585	0.612
	Without attention	0.641	0.506	0.489	0.554
	RNALoc-LM	0.680	0.607	0.596	0.629
miRNA	Without RNA-FM (one-hot)	0.839	0.776	0.760	0.801
	Without RNA-FM (word2vec)	0.867	0.846	0.858	0.852
	Without CNN	0.869	0.855	0.851	0.862
	Without Bi-LSTM	0.885	0.874	0.867	0.884
	Without attention	0.880	0.868	0.862	0.877
	RNALoc-LM	0.887	0.876	0.869	0.887
circRNA	Without RNA-FM (one-hot)	0.797	0.788	0.797	0.784
	Without RNA-FM (word2vec)	0.640	0.557	0.635	0.617
	Without CNN	0.820	0.815	0.816	0.814
	Without Bi-LSTM	0.812	0.805	0.809	0.803
	Without attention	0.800	0.794	0.796	0.793
	RNALoc-LM	0.826	0.820	0.824	0.818

RNA type	Variant model	ACC	Macro F1	Macro precision	Macro recall
lncRNA	Without RNA-FM (one-hot)	0.641	0.418	0.376	0.510
	Without RNA-FM (word2vec)	0.579	0.418	0.385	0.494
	Without CNN	0.660	0.579	0.576	0.608
	Without Bi-LSTM	0.674	0.571	0.585	0.612
	Without attention	0.641	0.506	0.489	0.554
	RNALoc-LM	0.680	0.607	0.596	0.629
miRNA	Without RNA-FM (one-hot)	0.839	0.776	0.760	0.801
	Without RNA-FM (word2vec)	0.867	0.846	0.858	0.852
	Without CNN	0.869	0.855	0.851	0.862
	Without Bi-LSTM	0.885	0.874	0.867	0.884
	Without attention	0.880	0.868	0.862	0.877
	RNALoc-LM	0.887	0.876	0.869	0.887
circRNA	Without RNA-FM (one-hot)	0.797	0.788	0.797	0.784
	Without RNA-FM (word2vec)	0.640	0.557	0.635	0.617
	Without CNN	0.820	0.815	0.816	0.814
	Without Bi-LSTM	0.812	0.805	0.809	0.803
	Without attention	0.800	0.794	0.796	0.793
	RNALoc-LM	0.826	0.820	0.824	0.818

The best performance values are highlighted in bold.

Table 3.

Open in new tab

Performance of RNALoc-LM and its variant models on independent test sets for three types of RNAs.^a

RNA type	Variant model	ACC	Macro F1	Macro precision	Macro recall
lncRNA	Without RNA-FM (one-hot)	0.641	0.418	0.376	0.510
	Without RNA-FM (word2vec)	0.579	0.418	0.385	0.494
	Without CNN	0.660	0.579	0.576	0.608
	Without Bi-LSTM	0.674	0.571	0.585	0.612
	Without attention	0.641	0.506	0.489	0.554
	RNALoc-LM	0.680	0.607	0.596	0.629
miRNA	Without RNA-FM (one-hot)	0.839	0.776	0.760	0.801
	Without RNA-FM (word2vec)	0.867	0.846	0.858	0.852
	Without CNN	0.869	0.855	0.851	0.862
	Without Bi-LSTM	0.885	0.874	0.867	0.884
	Without attention	0.880	0.868	0.862	0.877
	RNALoc-LM	0.887	0.876	0.869	0.887
circRNA	Without RNA-FM (one-hot)	0.797	0.788	0.797	0.784
	Without RNA-FM (word2vec)	0.640	0.557	0.635	0.617
	Without CNN	0.820	0.815	0.816	0.814
	Without Bi-LSTM	0.812	0.805	0.809	0.803
	Without attention	0.800	0.794	0.796	0.793
	RNALoc-LM	0.826	0.820	0.824	0.818

RNA type	Variant model	ACC	Macro F1	Macro precision	Macro recall
lncRNA	Without RNA-FM (one-hot)	0.641	0.418	0.376	0.510
	Without RNA-FM (word2vec)	0.579	0.418	0.385	0.494
	Without CNN	0.660	0.579	0.576	0.608
	Without Bi-LSTM	0.674	0.571	0.585	0.612
	Without attention	0.641	0.506	0.489	0.554
	RNALoc-LM	0.680	0.607	0.596	0.629
miRNA	Without RNA-FM (one-hot)	0.839	0.776	0.760	0.801
	Without RNA-FM (word2vec)	0.867	0.846	0.858	0.852
	Without CNN	0.869	0.855	0.851	0.862
	Without Bi-LSTM	0.885	0.874	0.867	0.884
	Without attention	0.880	0.868	0.862	0.877
	RNALoc-LM	0.887	0.876	0.869	0.887
circRNA	Without RNA-FM (one-hot)	0.797	0.788	0.797	0.784
	Without RNA-FM (word2vec)	0.640	0.557	0.635	0.617
	Without CNN	0.820	0.815	0.816	0.814
	Without Bi-LSTM	0.812	0.805	0.809	0.803
	Without attention	0.800	0.794	0.796	0.793
	RNALoc-LM	0.826	0.820	0.824	0.818

The best performance values are highlighted in bold.