• My first RNA 3-D structure
• Examples from the litterature
• RNA Decoys database
• Why don't I obtain models!
• Modeling long hairpin loops
• Modeling internal loops
• Modeling multi-branched loops
• Modeling Bridging loops
• Modeling base triples
• Modeling with distance constraints
• Modeling base pair types
• Modeling coaxial stacking
• Modeling a pseudoknot
• Modeling with relations
• Modeling classic secondary structures
• Modeling with NCM sets
• Modeling with fragments
• Modeling a tRNA
• Sampling an RNA
• Modeling large RNAs
• Modeling with NMR restraints
• Adding fragments to a partial structure
• Using MC-Sym locally

References & Links:

• Parisien M, Major F. Nature. (2008) 452:51-55. (pubmed) (Nature)
• The MC-Sym web page
• Dr. Francois Major's email: major at iro.umontreal.ca
• Marc Parisien's email: parisien at iro.umontreal.ca

• Complete models [script] [workdir]
• Internal loop hypothesis #1 [script] [workdir]
• Internal loop hypothesis #2 [script] [workdir]
• Internal loop hypothesis #3 [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

• Complete models from observed NCMs [script] [workdir]
• Complete models from artificial NCMs [script] [workdir]
• Secondary structure [.eps]

Step 1. Confirm that MC-Sym is running.

000 (7355.000.000) 11/25 00:57:11 Job submitted from host: <10.0.0.156:10054>
...
001 (7355.000.000) 11/25 00:57:14 Job executing on host: <10.0.0.233:11011>
...

Step 2. Locate the `Build Order` section.

> Build order (4.27595e+09 potent solutions):
[0  ]: Place[ncm_01] {10}
[1  ]: Merge[ncm_02]+{[A7|R:RIB],[A10|R:RIB]} {1}
[2  ]: Merge[ncm_03]+{[A6|R:RIB],[A11|R:RIB]} {1}
...

Step 3. Locate the last `Reached` section.

Reached level 0 Place[ncm_01]

Step 4. Inspect the `Building FG` section.

===== Building FG "ncm_02" =====

o            Reading 21 PDB files into LibraryFG "ncm_02"

o 1 models added into library "ncm_02" (read 53).

Step 5. Edit your MC-Sym script.

ncm_02 = library(
	pdb( "..." ) ...
	rmsd( 1.0 sidechain && !( pse || lp || hydrogen ) ) )

ncm_02 = library(
	pdb( "..." ) ...
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

Step 6. Cancel your current job and re-submit.

• MC-Sym script

Step 1: Identify the fragments to use.

Step 2: Load the fragments.

lnk_01 = library(
	pdb( "MCSYM-DB/ss3/CUU/*.pdb.gz" ) #1:#3 <- A31:A33
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

lnk_02 = library(
	pdb( "MCSYM-DB/ss4/CGCG/*.pdb.gz" ) #1:#4 <- A36:A39
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

lnk_03 = library(
	pdb( "MCSYM-DB/ss3/GUC/*.pdb.gz" ) #1:#3 <- A34:A36
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

Step 3: Use the fragments.

//----- stem -----
        ....      

//----- loop -----
merge( lnk_01 2.0 )
merge( lnk_02 2.0 )
merge( lnk_03 2.0 )

References:

1. Merino et al. J Am Chem Soc. 2005 127:4223-31.
2. Durant & Davis. J Mol Biol. 1999 285:115-31.
3. Perret et al. Biochimie. 1990 72:735-43.

• MC-Sym script

Step 1: Identify the fragments to use.

Step 2: Load the fragments.

lnk_01 = library(
	pdb( "MCSYM-DB/ss2/CU/*.pdb.gz" ) #1:#2 <- A23:A24
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

lnk_02 = library(
	pdb( "MCSYM-DB/ss4/AUCU/*.pdb.gz" ) #1:#4 <- A6:A9
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

Step 3: Use the fragments.

//----- upper stem -----
           ...

//----- interior loop -----
merge( lnk_01 2.0 )
merge( ncm_06 2.0 )
merge( lnk_02 2.0 )

//----- lower stem -----
           ...

References:

1. Aboul-ela et al. Nucleic Acids Res. 1996 24:3974-81.
2. Aboul-ela et al. J Mol Biol. 1995 253:313-32.

• MC-Sym script

Step 1: Identify the fragments to use.

Step 2: Load the fragments.

lnk_01 = library(
	pdb( "MCSYM-DB/ss2/AA/*.pdb.gz" ) #1:#2 <- A25:A26
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

lnk_02 = library(
	pdb( "MCSYM-DB/ss2/UC/*.pdb.gz" ) #1:#2 <- A37:A38
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

lnk_03 = library(
	pdb( "MCSYM-DB/ss4/CUGA/*.pdb.gz" ) #1:#4 <- A6:A9
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

Step 3: Use the fragments.

//----- multi-brached loop -----
ncm_07
merge( lnk_01 2.0 )
merge( ncm_12 2.0 )
merge( lnk_02 2.0 )
merge( ncm_13 2.0 )
merge( lnk_03 2.0 )

//----- stem_1 -----
           ...

//----- stem_2 -----
           ...

//----- stem_3 -----
           ...

References:

1. Lescoute & Westhof. RNA. 2006 12:83-93.
2. Tyagi & Mathews. RNA. 2007 13:939-51.
3. Diamond et al. Biochemistry. 2001 406971-81.
4. Dunham et al. J Mol Biol. 2003 332:327-36.
5. Kruger et al. Cell. 1982 31:147-57.
6. Guerrier-Takada et al. Cell. 1983 35:849-57.

References:

1. Cate et al. Science. 1996 273:1678-85.

• MC-Sym script

Step 1: Build the triple as an internal loop.

Step 2: Enforce the pairings to form the triple.

implicit_relation(
  // these constraints will define the base triple
  A7 A23 { pairing }
  A8 A23 { pairing }
)

References:

1. Chowdhury et al. Embo J. 2006 25:2487-97.

Inferred distance constraints are cheap as they imply no further experimental procedures. Consider the GNRA tetraloop. Find it's receptor [1,2] in the structure you're modeling, and you'll have an additional distance constraint. Comparative sequence analysis can also unveil additional inter-helical or inter-domain base pairs, which can in turn serve as distance constraints [3].

Experimentally derived distance constraints can be obtained from high resolution methods, like NMR or X-Ray, but also from mid- and low-resolution methods like cryo-EM, SAXS, and multiplexed hydroxyl radical cleavage analysis (MOHCA) [4].

Consider the data amassed by Das et al. using MOHCA [4], which can be readily used in MC-Sym as distance constraints. The hydroxyl cleavage traces are between sugars less than 25 Angstroms apart. In MC-Sym, since the distance constraints are hard (i.e. cannot be violated) we can extend this range to, say, 30 Angstroms. Hence, for each pair of nucleotides picked up by the method we can add a constraint in the MC-Sym script to reflect the proximity of the pair (see Table S2 of [4]):

distance( A109:C4'  A182:C4'  4.0  30.0 ) // P4_P5 / P5A
distance( A113:C4'  A197:C4'  4.0  30.0 ) // P4_P5 / P5A
distance( A114:C4'  A196:C4'  4.0  30.0 ) // P4_P5 / P5A
distance( A115:C4'  A195:C4'  4.0  30.0 ) // P4_P5 / P5A
distance( A121:C4'  A193:C4'  4.0  30.0 ) // P4_P5 / P5A

Distance constraints can be also used for nucleobase stacking:

distance( A25:PSY  A26:PSY  0.0  5.0 )

Distance constraints can be also used for nucleobase pairing; typical C1'-C1' distances for W/W pairs are:

• 11 Angstroms, for purine + pyrimidine
• 9 Angstroms, for pyrimidine + pyrimidine
• 14 Angstroms, for purine + purine

Finally, distance constraints can steer MC-Sym in large loop modeling or serve as gards for unmodeled parts of an RNA. Indeed, for two nucleotides i and i+k the 99th percentile distance D between their C1' atoms goes like:

// D = 4.4 * k + 5.8
distance( B40:C1'  B43:C1'  0.0   19.0 )

References:

1. Jaeger et al. J Mol Biol. 1994 236:1271-6.
2. Cate et al. Science. 1996 273:1678-85.
3. Massire et al. J Mol Biol. 1998 279:773-93.
4. Das et al. PNAS. 2008 105:4144-9.

implicit_relation(

  // the G=A base pair flanking a GNRA tetraloop
  // in the Leontis/Westhof style
  A8 A23 { S/H && antiparallel && trans }

  // the canonical C=G Watson-Crick base pair
  // in the Saenger style
  A7 A24 { XIX }

  // the most underspecified pair
  A6 A25 { pairing }
)

• The relative glycosidic bonds orientation in a base pair can be either cis or trans
• The Saenger notation describes base pairs with at least two hydrogen bonds.
  Valid pairs are I to XXVIII (table), but notably:
• XIX: the C=G Watson-Crick
• XX: the A=U Watson-Crick
• XXVIII: the G=U Watson-Crick
• The faces implicated in a base pair can be one of:
  W S H Ws Ww Wh Sw Ss Hw Hh C8 Bs Bh O2P O2'
• The parallel or antiparallel describes the relative strands orientation.

References:

1. Saenger. Principles of nucleic acid structure. 1984.
2. Leontis & Westhof. RNA. 2001 7:499-512.
3. Lemieux & Major. Nucleic Acids Res. 2002 30:4250-63.

distance( A25:PSY  A26:PSY  0.0  5.0 )

• MC-Sym script

References:

1. Parisien & Major. Nature. 2008 452:51-5.
2. Mathews et al. PNAS. 2004 101:7287-92.
3. Wu & Feigon. PNAS. 2007 104:6655-60.

• MC-Sym script

References:

1. Aalberts & Hodas. Nucleic Acids Res. 2005 33:2210-4.
2. Holland et al. RNA. 1999 5:257-71.

• MC-Sym script

Step 1: Declare the relations.

relation(
  A9  A10 { adjacent_5p } 75%
  A10 A23 { pairing }     75%
  A9  A25 { pairing }     75%
  A25 A24 { pairing }     75%
)

relation(
  A9  A10 { adjacent_5p && !stack } 25%
)

relation(
  A74:A76 { file( "helixA_RNA" ) stack } 1
  A15 A48 { file( "helixA_RNA" )  W/W  } 1
)

Step 2: Use the relations.

structure = backtrack(
    // the triple
    ( A9 A25 A24 )
    // next stage
    ( A9 A10 )
    // the other base pair
    ( A10 A23 )
)

References:

1. Cevec et al. Nucleic Acids Res. 2008 36:2330-7.

Classic secondary structures can be further analyzed by MC-Fold for it to produce enhanced secondary structures. These now could then be modeled in 3-D more easily using MC-Sym.

References:

1. Correll et al. PNAS. 1998 95:13436-41.
2. Szewczak et al. PNAS. 1993 90:9581-5.
3. Wimberly et al. Biochemistry 1993 32:1078-87.
4. Das & Baker. PNAS. 2007 104:14664-9.
5. Martinez et al. J Biomol Struct Dyn. 2008 25:669-84.
6. Ding et al. RNA. 2008 14:1164-73.

• *_t: theoretical set, only for 2_2 NCMs with A=U/C=G base pairs. For example:

  ncm_01 = library( pdb( "MCSYM-DB/2_2/UCGA/*_t.pdb.gz" ) ... )
  

  It's purpose is to provide only one 3-D instance for that NCM, reducing the conformational search space. Useful to model large RNAs, or to force MC-Sym to sample other parts of the RNA structure. One can use the R20 sets instead (see below); which would also add the G=U base pair in a strict set with few, but "high quality", NCMs. Two problems arise when using the _t sets; 1) consecutive pyrimidines won't be stacked, at least on the 5' strand. 2) base pairs won't be propelled.



• *_1: observed set, found in the PDB with the actual sequence. For example:

  ncm_01 = library( pdb( "MCSYM-DB/4/GGAA/*_1.pdb.gz" ) ... )
  

  It's purpose is to provide 3-D instances for that NCM that can be found in the PDB with the actual modeled sequence. Useful to model parts of an RNA structure which are known to adopt particular folds, like the GNRA-tetraloops, or base pair steps with proper helical parameters [1,2,3].



• *_x: artificial set, built from backbone templates. For example:

  ncm_01 = library( pdb( "MCSYM-DB/4_2/GUUCGC/*_x.pdb.gz" ) ... )
  

  It's purpose is to provide 3-D instances for that NCM that were built from a database of backbone templates onto which the flanking base pairs were fitted, and any other nucleobases added. Useful to model any NCMs which cannot be found in the PDB. As more nucleotides are implicated in the NCM, the number of backbone templates diminishes rapidly. Hence, a user should opt for other modeling options, like modeling internal loops. The artificial set construction is described in details in [4].



• *_e: energy optimized artificial set, built from backbone templates. For example:

  ncm_01 = library( pdb( "MCSYM-DB/2_2/GAGA/*_e.pdb.gz" ) ... )
  

  This set is a subset of *_x, in which each NCM has been annotated, and only the ones with best energies are kept, for each suite of base pairs. This set is only available for 2_2 NCMs. This set is useful to sample many base pair types, but with few instances only (to keep the search space at minimum).



• *_R*: 2_2 NCMs that come from X-Ray only. For example:

  ncm_01 = library( pdb( "MCSYM-DB/2_2/GAGA/*_R*_1.pdb.gz" ) ... )
  

  This set is a subset of *_1, in which each NCM has been solved with X-Ray. This set is only available for 2_2 NCMs.



• *_Rr*: 2_2 NCMs that come from X-Ray only, with a resolution better or equal to r. For example:

  ncm_01 = library( pdb( "MCSYM-DB/2_2/GAGA/*_R20*_1.pdb.gz" ) ... )
  

  This set is a subset of *_R, in which each NCM has been solved with X-Ray, at a resolution better or equal to r. This set is only available for 2_2 NCMs. Here, R20 are those NCMs from resolutions better or equal to 2.0A. Available sets are R10, R15, R20, R25, R30 and R35, for resolutions better or equal to, respectively, 1.0, 1.5, 2.0, 2.5, 3.0 and 3.5 Angstroms. Hence, by using R35, you also use those lower than 3.5A, that is R30 downto R10. Theoretical NCMs featuring stacks of Watson-Crick base pairs could be replaced with the R20 sets, for example. Don't expect to find your favorite 2_2 NCM in the R10 set though (i.e. solved at a resolution better than 1.0 Angstrom)!



• *: a combination of all other sets. For example:

  ncm_01 = library( pdb( "MCSYM-DB/4/UACG/*.pdb.gz" ) ... )
  

  However, MC-Sym will load in that order: 1) the theoretical set (if any), 2) the observed set (if any), and 3) the artificial set.

References:

1. Hunter. J Mol Biol. 1993 230:1025-54.
2. Hunter. J Mol Biol. 1998 280:407-20.
3. Hunter & Lu. J Mol Biol. 1997 265:603-19.
4. Parisien & Major. Nature. 2008 452:51-55. Supp Inf

triple = library(
	pdb( "../IXYuDpUiAB/*.pdb.gz" ) #1:#2, #3:#5 <- A9:A10, A23:A25
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

triple = library(
	pdb( "../{WorkDir1,WorkDir2,...}/*.pdb.gz" ) #1:#2, #3:#5 <- A9:A10, A23:A25
	rmsd( 0.1 sidechain && !( pse || lp || hydrogen ) ) )

• The MC-Sym script
• The canonized tRNA 3-D structure, without modified nucleotides
• Superimposed MC-Sym solution (model 2) on crystal (model 1)

References:

1. Major et al. PNAS. 1993 90:9408-12.
2. Jovine et al. J Mol Biol. 2000 301:401-14.

• The MC-Sym script which samples the acceptor stem
• The MC-Sym script which samples the anticodon stem/loop
• The MC-Sym script which samples the T stem/loop
• The MC-Sym script which samples the D stem/loop
• The MC-Sym script which assembles the whole tRNA using previous samples

References:

1. Grishaev et al. J Biomol Nmr. 2008 42:99-109.

1. Model each stem/hairpin separately.
• Generate many MC-Sym models (change model_limit = 9999).
• use the SCORE + FILTER in the commands.html file,
  once all models are generated.
2. Model the whole structure using previously built stems/hairpins.
   Consider the Modeling with fragments section.

WARNING If you do not have distance constraints between your stems/hairpins then the number of possible 3-D structures is as large as there are grains of sand on a beach!

The other way to exploit the NMR data is to evaluate the fitness of each MC-Sym model with respect to the NMR restraints. Basically, you should:

1. Generate many MC-Sym models (change model_limit = 9999).
2. Use the "NMR" section in the "commands.html" file,
   once all models are generated.

Supported format #1:

    15  GUA H1             19  GUA H1             3.500    6.500  
    19  GUA H1             13  URA H3             2.600    5.000  
    20  ADE H2             12  GUA H1             2.600    5.000  
    20  ADE H2             13  URA H3             1.800    3.600  
    20  ADE H2             19  GUA H1             1.800    3.600  

123456789012345678901234567890123456789012345678901234567890123456
         1         2         3         4         5         6

Supported format #2:

assign (residue 1 and name H2') (residue 1 and name H3')  2.60  0.78  0.78
assign (residue 1 and name H1') (residue 1 and name H3')  3.20  0.96  0.96
assign (residue 1 and name H3') (residue 1 and name H8)   3.20  0.96  0.96
assign (residue 1 and name H2') (residue 1 and name H8)   4.00  1.20  1.20
assign (residue 1 and name H1') (residue 1 and name H8)   2.90  1.10  1.10

* Other formats may be supported by explicit demand to Dr. Francois Major.

Installation.

1. Download MC-Sym from here.
2. Install MC-Sym as prescribed.
3. Goto the pipeline page here.
4. Click on the "snapshot" button of the "What is MC-Sym" section.
5. Download your tarball "MCSYM-DB.tar.gz" on your computer.
6. Unzip and untar the tarball:

   gunzip MCSYM-DB.tar.gz; tar xvf MCSYM-DB.tar

cd ~/RNA/Modeling/Hammerhead/
ln -s MCSYM-DB ~/RNA/MC-Sym/MCSYM-DB/

ncm_01 = library(
      pdb( "MCSYM-DB/4/GGCC/*.pdb.gz" )  ...

cd ~/RNA/Modeling/Hammerhead/
~/RNA/MC-Sym/mcsym  ./Hammerhead.mcc

Updating your database.

Failed to expand pattern 'MCSYM-DB/4/UACG/*.pdb.gz': No such file or directory

1. Make a "fake" MC-Sym script with all the "library" commands which fails. For example:

   // these are my missing blocks:
   library( pdb( "MCSYM-DB/4/UACG/*.pdb.gz" ) )
   library( pdb( "MCSYM-DB/4/UCCG/*.pdb.gz" ) )
   library( pdb( "MCSYM-DB/4/UGCG/*.pdb.gz" ) )
   library( pdb( "MCSYM-DB/4/UUCG/*.pdb.gz" ) )
   

2. Submit this "fake" script to MC-Sym here; wait until the job is queued.
3. Refresh your database; repeat steps 3 to 6 of the previous "Installation" section.

To go local or not?

• Automatic updates of the NCM database
• Scoring of the 3-D models with an RNA-tailored force-field
• Refinement of the backbones
• Distributed computing facillity
• Analysis of base pair types