Recommendations and Enhancements


Aria 1.2


This information is specific to ARIA version 1.2. I have no direct knowledge of latter versions of ARIA, but users of version 2 have indicated that at least part of this information is still pertinent.

ppm tolerances depend on data (example values are)
  • 1H -> 0.04
  • 15N -> 0.35
  • 13C -> 0.3
distance restraint force constants
  • unambig and ambig -> 0, 10, 15, 50, 100
  • hbonds -> 0, 10, 10, 50, 50
Spectra parameters
  • qrelax -> false

Structure refinement only using already calibrated data - no assignments (only does it0)
  • The following parameters will give results similar to that from CNS anneal.inp cartesian refinement
  • unambig.tbl, ambig.tbl, and dihedral force constants set to 50 (other schemes may also work)
  • structures = 50 (or whatever)
  • keep_structures set to how many you want to use for average (you may not know this value ahead of time)
  • use cartesian dynamics
  • initial temp for TAD and Cartesian = 2000
  • final temp 1st cooling = 1000
  • final temp 2nd cooling = 0
  • time step = .005
  • init # of steps = 10,000
  • # steps for 1st cooling = # steps for 2nd cooling = 5000
  • other values may work so don't be afraid to experiment

Aria on Macintosh (and some other platforms) cannot do torsional angle dynamics (see CNS discussion).
  • Below are modifications to the Aria protocols files needed to use TAD on Macintosh.
  • These files are in /usr/local/aria1.2/protocols
  • sa_ltad_hightemp4.cns
  • sa_ltad_cool1.cns
  • change tcoupling=true to vscaling=true in both of the above protocol files

ARIA program modification
  • This modification has been made on daffy and bugs standard and dimer Aria routines. You may need to implement a similar scheme if you use Aria on another computer.
  • The Problem - Aria will sometimes make an error while merging calibrated restraint files into the ambig.tbl and unambig.tbl files. It depends on the number of crosspeaks, maxn, and the phase of the moon. It appears to be a timing error possibly due to an attempt to read a restraint file before it has been completely written or closed.
  • Possible Solution - Modify the python script aria1.2/Aria/Main/MAria.py
  • Go here to see the code changes

Tip on starting ARIA refinement from a starting structure instead of random structure
  • Edit run.cns and set the number of "keep_structures" during it0 so that new random structures are not generated for every iteration. If you don't want any random structures then set "keep_structures" to the value of "structures".
  • Set "violation tolerance" to values such that noes are not reassigned to distand atome. For example, set to 10 or 5 for it0, and then even less for subsequent iterations.
  • Start a run with run.cns. Wait until after Aria generates the template file and hit Control-c when it completes this task. (Watch output text) Then copy the starting pdb file (must have been generated with Xplor or CNS so atom names and numbering corresponds with psf file) to begin/*_template.pdb
  • Now you can start aria again and it will use your template instead of the extended chain that it usually uses.


CNS


anneal.inp - the default values work well for simulated annealing from a previously determined structure using already calibrated restraint (.tbl) files. However, it is not sufficient for annealing from an extended chain. Try the following changes:
  • cartesian instead of TAD
  • temp = 2000 (hot and 1st cooling)
  • md steps = 10,000 with step size = .005
  • vdw scaling =0.1 and 4.0 for hot and 1st cooling
  • vdw scaling for 2nd cooling starts at 1.0 and ends at 4.0
  • temp for 2nd cooling = 1000
  • noe and dihedral force constants = 50 (can try different values)
  • temperature cooling step = 25

accept.inp
  • may need to double default cutoffs for bonds and angles

Residual Dipolar Coupling
  • Must define atoms X, Y, Z, OO for defining the anisotropy tensor. Include these 3 files provided with Aria.
  • include /usr/local/aria1.2/toppar/tensor.pdb as a second coord file to anneal.inp
  • include /usr/local/aria1.2/toppar/tensor.psf as a second structure file to anneal.inp
  • include /usr/local/aria1.2/toppar/par_axis.pro as a second topology file to anneal.inp
  • Enter name of formatted RDC file in SANI section
  • Set DFS = 0
  • Calculate axial and R from Szz, Syy, and Sxx

CNS on Macintosh (and some other platforms)
  • Select velocity rescaling instead of Berendson.
  • CNS on Macintosh cannot do torsional angle simulared anealing using the default anneal.inp task file. Apparently Berendsen's temperature coupling fails. You must change the protocol to use velocity rescaling instead.
  • anneal.inp has been modified to include this option. Only the version on the web server needs to be modified. Our daffy server already has this done.
  • click here for changes to anneal.inp described above
  • the other annealing task files may not have this option applied.


Aqua 3.2

conversion of xplor restraint files from Aria does not work properly.
  • modify these two files
  • aqua3.2/scripts/convXPLORdistrestr
  • aqua3.2/scripts/qmodr
  • this may affect conversion of other file types
  • see here for code changes


Python


download and install Python 2.3 as directed, except you must use the following option with make when compiling
  • make OPT=-O