2.7 | RNA structure
For a number of years, RNA Puzzles (6) has evaluated RNA modeling
methods on a rolling basis as experimental structures became available.
The rate of RNA experimental structure determination has increased
markedly recently, partly because of advances in cryo-electron
microscopy. Consequently, it is now possible to obtain enough targets to
make a CASP category for this area possible. The CASP15 RNA category was
conducted in close collaboration with RNA Puzzles, and assessed by joint
team led by Rhiju Das (for CASP) and Eric Westhof (for RNA Puzzles)
(13). There were 12 targets in all, ranging in size from 30 bases to 720
bases, and including four designed structures. Two of the structures are
of the same RNA molecule with a different number of partner proteins
bound.
The assessors judged that except for the two RNA-protein complexes, the
overall topology was correct for at least some models of each target,
and Watson-Crick base pairing was also largely accurate, as was the
relative positioning of the double helical regions. However,
non-canonical pairing and irregular conformations regions were less
successfully modeled. They also judged that most targets display
evidence of significant flexibility, complicating assessment. For one
target, the experimental group generated four different cryo-EM maps.
The highest agreement of any submitted model was to one of the secondary
density maps (Figure 7), illustrating that for some RNA structures
additional experimental work may be needed for a proper assessment of
computational models.