CYANA
Introduction
CYANA is a macromolecular structure calculation algorithm based on simulated annealing molecular dynamics calculations in torsional angle space, in contrast to Cartesian space (Ref. 1,2). Here the only degrees of freedom are torsion angles with covalent structure parameters kept fixed, thereby significantly decreasing the number of degrees of freedom in the calculation.
The current version of CYANA is 3.0, and it is capable of handling orientational (i.e., RDC) constraints. There is now a nice CYANA 3.0 wiki with file explanations, tutorials and theory.
In the sections below we describe input files and protocols used in the NESG for both CYANA 2.1 and 3.0.
CYANA 2.1
Residue Library
CYANA 2.1 uses a new library ~/lib/cyana.lib. It is loaded by the cyanalib command. For back-compatibility there is also the old DYANA library dyana.lib (loaded with dyanalib, of course).
The main difference is larger van der Waals radii. This will give you a larger target function than DYANA, but also better clash scores.
Residue nomenclature is also different - there are no charged species, like ARG+ or GLU-.
Atom Nomenclature
Atom nomenclature was made compatible with BMRB standard. The deviations from XEASY/DYANA conventions are: HN <-> H, HA1 <-> HA2 and HA2 <-> HA3 for GLY.
There a is macro translate.cya, which is used to convert input to different formats. For example, to read files with DYANA nomenclature, enter translate dyana. To switch back to CYANA 2.1 convention type translate off
Pseudoatom Treatment
Pseudoatom handling is switched by setting pseudo=x, where x is 0, 1, 2, or 3.
With pseudo=0, the default setting, coordinate files *.cor and *.pdb do not contain pseudoatoms. They are calculated implicitly on the run.
Setting pseudo=1 restores the old DYANA behavior with explicit pseudoatoms.
Setting pseudo=2 switches to simplified pseudoatom names, such as HB instead of QB, HD1 instead of QD1, and HD instead of QQD of Leu. This is the setting to be used when reading chemical shifts from CARA. Coordinate files will contain explicit pseudoatoms, as with pseudo=1
Setting pseudo=3 allows X-Plor/CNS pseudoatom names, like HX* instead of QX. For some reason using translate xplor is not enough to do the conversion for all the atoms.
CALC.cya and Structure Calculation
There are calculation demos for automatic assignment (~/demo/auto) and simple structure calculation (~/demo/manual) runs.
CALC.cya script for automatic NOE assignment:
peaks := c13.peaks,n15.peaks,aro.peaks # names of NOESY peak lists prot := demo # names of chemical shift lists constraints := demo.aco # additional (non-NOE) constraints tolerance := 0.040,0.030,0.45 # chemical shift tolerances calibration := # NOE calibration parameters structures := 100,20 # number of initial, final structures steps := 10000 # number of torsion angle dynamics steps rmsdrange := 10..100 # residue range for RMSD calculation randomseed := 434726 # random number generator seed noeassign peaks=$peaks prot=$prot autoaco
To prevent CYANA from changing existing peak assignments you need to define a subroutine to select the peaks to keep:
peaks := c13.peaks,n15.peaks,aro.peaks # names of NOESY peak lists prot := demo # names of chemical shift lists constraints := demo.aco # additional (non-NOE) constraints tolerance := 0.040,0.030,0.45 # chemical shift tolerances calibration := # NOE calibration parameters structures := 100,20 # number of initial, final structures steps := 10000 # number of torsion angle dynamics steps rmsdrange := 10..100 # residue range for RMSD calculation randomseed := 434726 # random number generator seed subroutine KEEP peaks select "*, * number=2..7999" end noeassign peaks=$peaks prot=$prot autoaco keep=KEEP
Here, subroutine KEEP is used to keep the assignments for peaks with peak numbers from 2 to 7999.
CALC.cya script for manual structure calculation:
peaks := c13,n15,aro # names of peak lists
prot := demo # names of proton lists
tolerance := 0.040,0.030,0.45 # chemical shift tolerances
# order: 1H(a), 1H(b), 13C/15N(b), 13C/15N(a)
calibration:= 6.7E5,8.2E5,8.0E4 # calibration constants (will be determined
# automatically, if commented out)
dref := 4.2 # average upper distance limit for
# automatic calibration
if (master) then
# ---- check consistency of peak and chemical shift lists----
peakcheck peaks=$peaks prot=$prot
# ---- calibration ----
calibration prot=$prot peaks=$peaks constant=$calibration dref=$dref
peaks calibrate "**" simple
write upl $name-in.upl
distance modify
write upl $name.upl
end if
synchronize
# ---- structure calculation ----
read seq $name.seq # re-read sequence to initialize
read upl $name.upl # read upper distance limits
read aco $name.aco # read angle constraints
seed=5671 # random number generator seed
calc_all structures=100 command=anneal steps=10000 # calculate 100 conformers
overview $name.ovw structures=20 pdb # write overview file and coordinates
Note the order in which tolerances are given.
The calibration field can be left empty, in this case dref will be used to derive calibration constants. If dref is not specified noeassign.cya will use a default value of 4.0. During calculation noeassign.cya will also relax the calibration if needed (that is in "elastic" mode, which is the default).
constraints need not be non-NOE despite what the comment says. You can add *.aco, *.upl, *.lol, and even *.cya macros for stereospecific assignments (haven't tested it yet, but that's the way CYANA adds stereospecific assignments in the final round).
master and synchronize keywords are needed for running on a cluster.
peakcheck checks the peaklist assignments against the atom list. Always check CYANA output for peakcheck results - those huge upl violations may be caused by mis-assigned peaks.
NOE Calibration
Here arguably lies the most confusing change since DYANA / CYANA 1.0. The current CYANA 2.1 by default does not use explicit pseudoatom corrections in distance constraints. Instead, these corrections are applied implicitly on-the-fly. This behavior is turned on by setting expand=1.
Calibration is thus performed with the undocumented statement peaks calibrate "**" simple. Trivial calculations show, however, that this command uses a simple r^-6 calibration without adding pseudoatom corrections.
Old calibration macros, such as calibrate.cya and caliba.cya are still allowed, but they do add explicit pseudoatom corrections. So if want to use them, don't forget to set expand=0. Omitting it will result in applying corrections twice, making the corresponding constraints very loose.
This is, of course, a matter of huge confusion since both methods produce otherwise identical *.upl files. Be sure you know HOW you calibrate you NOEs.
To modify upper and lower distance cutoffs for NOE calibration, use set upl_values:=2.4,6.0. The defaults are 2.4 and 5.5.
Stereospecific assignments
Constraints for diastereotopic atoms (such as HB2, HB3) are treated as ambiguous by CYANA. This is switched on with swap=1.
For the manual run you may want to have swap=0 to be compatible with DYANA behavior. This option is apparently not necessary when distance modification is applied.
Distance modification does not affect Phe and Tyr ring atoms HD1/2 and HE1/2. Therefore, if you have degenerate ring chemical shift (as is almost always the case) make sure you have them labeled QD and QE
External stereospecific assignments determined with glomsa or with the help of a 5% 13C-labeled sample (Biochemistry 1989 28(19) p7510) can be defined with a custom macro like this:
# VAL atom stereo "QG1 25 36 38 87" atom stereo "QG1 43 90" atom swap "QG1 43 90" # LEU atom stereo "QD1 60 63 97" atom stereo "QD1 35 56" atom swap "QD1 35 56"
Here the syntax of CYANA 2.1 requires double quotes. For some strange reason, methyl groups should be written with the letter "Q" even if pseudo=2 is used.
Peaklists
CYANA 2.1 can produce multiple assignments for a peak. Below is a part of an aliphatic NOESY peaklist with peak #6 having two assignments. #VC tags specify the weights given to individual assignments. Calibration of this peak yields two constraints splitting the peak integral according to these weights.
# Number of dimensions 3
#FORMAT xeasy3D
#INAME 1 H
#INAME 2 C
#INAME 3 h
#CYANAFORMAT HCh
1 4.147 51.731 1.474 3 U 7.953E+03 0.000E+00 - 0 2234 2233 2238 #QU 1.000 #SUP 1.00
2 4.147 51.731 4.251 4 U 4.181E+03 0.000E+00 - 0 0 0 0
3 4.147 51.731 7.791 3 U 6.017E+03 0.000E+00 - 0 2234 2233 232 #QU 1.000 #SUP 1.00
4 1.474 22.186 0.515 3 U 1.481E+03 0.000E+00 - 0 2390 2389 1417 #QU 0.981 #SUP 0.98
5 1.474 22.186 1.249 3 U 2.610E+04 0.000E+00 - 0 2390 2389 1706 #QU 0.987 #SUP 0.99
6 1.474 22.186 2.635 3 U 1.396E+03 0.000E+00 - 0 2390 2389 1715 #VC 0.47897 #QU 0.774 #SUP 0.96
2390 2389 1815 #VC 0.52103 #QU 0.813 #SUP 0.96
7 1.474 22.186 3.863 3 U 1.418E+04 0.000E+00 - 0 2390 2389 1657 #QU 0.885 #SUP 0.88
8 1.474 22.186 4.147 3 U 1.448E+04 0.000E+00 - 0 2238 2237 2234 #QU 1.000 #SUP 1.00
The peaklists produced by CYANA 2.1 are not backwards-compatible with XEASY, but there are Lua scripts, which can read them into CARA including the information on ambiguous assignments. UBNMR should also be able to handle them in the future.
When supplying completely unassigned peaks for automatic NOE assignment it is necessary to include a line like #CYANAFORMAT HCh in the header.
05 Jul 2007 - 21:55 by AlexEletski