Stepwise Monte Carlo Server Documentation


Stepwise Monte Carlo is a tool for building a 'full model' of a complex RNA motif. Its central idea is that even complex motifs can be assembled by a Monte Carlo scheme that samples single nucleotides individually, one by one, with explicit additions and deletions. Input to the 'stepwise' algorithm consists of a description of the target motif -- provided as a FASTA file -- as well as input 'chunks' of RNA, provided as PDB files, as well as -- optionally -- a native PDB, if one exists, to evaluate how well the native conformation was recovered.


  1. A single input chunk does not have to be all contiguous: a single PDB may contain a single base pair or two from two different helices. That tells stepwise to respect the input rigid body orientation between the input structures and to sample only conformations that exactly close those particular orientations. In contrast, providing multiple PDBs ensures that rigid body degrees of freedom between them will be sampled.

  2. You do not have to either provide input chunk PDBs or a dot-bracket secondary structure as inputs. You can combine the two -- for example, if you want to supply a distorted crystal helix in addition to an ideal model generated by Rosetta. Importantly, though, our code will automatically generate all helices found in the secondary structure string, and they will be included in the stepwise command line even if they overlap, likely causing the job to fail. If you are providing some Watson-Crick helices as chunks, you should turn their corresponding secondary structure characters into '.'s. This won't cause any trouble to the run, because we don't use the secondary structure at all except to generate initial ideal helical chunks.
  3. While SWM is usually capable of making accurate binding predictions, some RNAs are very difficult to study, the following guidelines can help maximize the likelihood of obtaining high quality predictions:

    1. Problems containing multiple input chunks can be very challenging to solve entirely ab initio. A useful middle ground is to provide some part of the native geometry as an 'align_pdb': stepwise will require that conformations fall within 4.0 A of the input atoms. (For example, in a three way junction where one provides two-bp helices for each input chunk, one may wish to supply two of those two-bp helices in their native orientation as an 'align_pdb'.)

    2. You'll want to scale your number of cycles and decoys in proportion to the total of how many nucleotides you're building and how many free chunks you must add. In general, an eight nucleotide problem will take somewhat more than twice the computational power as a four nucleotide problem, and a 'free' or aligned problem two way junction will take about twice the effort as its 'fixed' equivalent.

Please cite the following article when referring to results from our ROSIE server:

  1. Watkins, A. M.; Geniesse, C.; Kladwang, W.; Zakrevsky, P.; Jaeger, L.; Das, R. "Blind prediction of noncanonical RNA structure at atomic accuracy. bioRxiv 223305; doi:

  2. Lyskov S, Chou FC, Conchúir SÓ, Der BS, Drew K, Kuroda D, Xu J, Weitzner BD, Renfrew PD, Sripakdeevong P, Borgo B, Havranek JJ, Kuhlman B, Kortemme T, Bonneau R, Gray JJ, Das R., "Serverification of Molecular Modeling Applications: The Rosetta Online Server That Includes Everyone (ROSIE)". PLoS One. 2013 May 22;8(5):e63906. doi: 10.1371/journal.pone.0063906. Print 2013. Link

We welcome scientific and technical comments on our server. For support please contact us at Rosetta Forums with any comments, questions or concerns.

Modeling tools developed by the Das Lab at Stanford University. The Rosie implementation was developed by Andrew Watkins and Sergey Lyskov.