Table 1:

Features and their corresponding information gain scores enabling the separation between high confident and low confident miRBase entries.

Feature	IG	Feature	IG	Feature	IG
mirbase_hpl	0.130	sl	0.106	%A++%G/sl	0.101
mwmF/hpl	0.116	clsp	0.105	%G++%A/sl	0.101
hpl	0.116	clep	0.105	%C++%U/hpl	0.101
%G++%U/hpl	0.109	#A++#C	0.104	%U++%C/hpl	0.101
%U++%G/hpl	0.109	#C++#A	0.104	ns/hpl	0.100
mwmF/sl	0.109	Tm/sl	0.103	saln	0.099
%G++%U/sl	0.107	%A++%G/hpl	0.102	nl/hpl	0.098
%U++%G/sl	0.107	%G++%A/hpl	0.102	#A++#G	0.098
Tm/hpl	0.106	#C++#U	0.102	#G++#A	0.098
#mdn	0.106	#U++#C	0.102	%C++%U/sl	0.098

For more information about the features please refer to Supplementary Table 2 and http://jlab.iyte.edu.tr/software/mirna/featureList. Bolded features are based on structure and thermodynamics while the rest is sequence-based. IG, Information gain.