Running ARP/wARP : The arp_warp.sh script.

This is the script that you will be running in most cases. If you execute it without any parameters, you get a short on-line help.

Define mode : warpNtrace / molrep /solvent n / warp
i.e.: arp_warp.sh mode warpNtrace for automatic model building ...

Possible arguments are: conservative / keepmodel / fsigma n / rsigma n / multitrace n / skip / notrunc /side
side : try to dock to sequence given in seq.pir
keepmodel: Do not discard model in first cycle
fsigma n: Define sigma level for finding new atoms
rsigma n: Define sigma level for removing atoms
skip: Skip first building cycle to refine first
conservative: Do not remove already build model parts during ARP refinement
update n: Update n* atoms than suggested by the program
notrunc: Do not truncate shifts greater than 3*rms
multitrace n: Try to autotrace the map more that once each time

The command line arguments you supply are common for all modes and can be given in any order! Here follows a short description for what they do and some hints for when to use them and when not. Note that all these parameters have reasonable default values, so only start fiddling with them after you understand what they do.

• side: This can use side chain building during cycling. Not much tested but technically it works.
• keepmodel: It's only relevant to warpNtrace. Normally warpNtrace just treats all atoms as free in the beginning and tries to see by itself later what's protein and what's not ... this one tells it to use the geometry implied in the start model and do not reset types to free atoms. Useful mostly in combination with skip (see below).
• fsigma n: Sigma threshold for finding atoms, if you want to reset it. We are trying to find a fool-proof way to get better thresholds, but sometimes if not enough atoms are removed or added you should play with it.
• rsigma n: Sigma threshold for removing atoms.
• skip: With this argument you force the script to skip the first autobuilding for a big cycle and only later re-autobuild. Its mostly useful with keepmodel if you start from a molecular replacement model or a partially built model. Using this argument you let this model refine a bit and use the improved phases for the first autotracing attempt. On it's own it can be used to do a 'big' cycle of refinement of free atoms and only then autobuild.
• conservative: This is really relevant only to warpNtrace. By default protein atoms as well as free atoms can be removed if they look bad to ARP for any reason. This way you can force ARP to keep protein atoms of the model while cycling.
• update n: The scripts calculate how many atoms to add or remove per cycle based on model completeness and data. If you think you know better you can give a multiplier factor ... i.e. 'update 2' will add and remove two times as many atoms as the script calculates.
• notrunc: Since the authors are exceptionally stubborn they wrote a little program (T_SHIFT)that resets excessive shifts for some atoms. Works as the RMSHIFT card in SFALL. For example, if some shifts are more than three times the rms shift the shift vectors for these atoms are truncated to three rms. If the rms shift is more than 0.5 Å then all shift vectors are truncated.
• multitrace n: Sometimes, if you first do an autotrace and use the output pdb file (which as some atoms better geometrised after real space fit of peptides) for one more cycle of tracing, you can gain a bit ... nothing dramatic really and not thoroughly tested either ...

Ok, now we will consider the different modes of the arp_warp.sh script operation.