Table 1.

RNA molecules included in the dataset

Molecule	PDB	l seq	S1–S4
yeast Phe-tRNA	1EHZ	76	VTTT
D5,6 Yeast ai5g G-II Intron	1KXK	70	TTTT
Ribonuclease P RNA	1NBS	150	TTTT
Adenine riboswitch	1Y26	71	VTTV
TPP riboswtich	2GDI	78	TTVT
SAM-I riboswitch	2GIS	94	TTVT
Lysine riboswitch	3DIG	174	TTVV
O. I. G-II Intron	3IGI	388	TVTV
c-di-GMP riboswitch	3IRW	90	TTTT
M-box riboswitch	3PDR	161	VVTT
THF riboswitch	3SD3	89	VVTT
Fluoride riboswitch	3VRS	52	VTTT
SAM-I/IV riboswitch	4L81	96	TVVV
Lariat capping ribozyme	4P8Z	188	TTTT
ydaO riboswitch	4QLM	108	TTTT
ZMP riboswitch	4XW7	64	VVVV
50S ribosomal	4YBB_CB	120	TVVV
5-HTP RNA aptamer	5KPY	71	TTTT

For each molecule we indicate the PDB ID of the corresponding annotated structure, the number of nucleotides (l_seq), and, for each random dataset splitting that we used (S1–S4), a mark to denote whether the molecule data are used for training (T) or validation (V). For PDB 4YBB, chain CB was used as a reference.