Hi Vikram,
Try Buccaneer, it works much better at such resolution than Arp/Warp.
Launch it several times with different parameters (e.g. use original
structure as seed/initial/nothing), then open the results together and
merge the pieces that look best and have lowest b-factors. Use Buster
for refinement of the merged model and iterate with Buccaneer again.
That's what worked for me once to bootstrap a poor MR solution (~20% of
the structure originally) to a near-complete model at 2.7A.
You are in a sense replicating what phenix.autobuild attempts to do, but
you'll be able to do it better (and probably faster) manually.
Kind regards,
Dmytro.
On 25/10/16 14:16, Vikram Dalal wrote:
Hi everyone,
We are trying to solve a protein structure of 2.6 A. We have processed
it with HKL2000. We have even tried processing with mosflm and xia2.
It is in C2221 space group (checked by pointless) and data is not
twinned. It has 31% identical with a search model and has 57% sequence
coverage. There are 2 subunits in asu.
I did not get proper phases with MOLREP and phaser. I have tried
Balbes (R free 50) and Mr BUMP (R free 54).But, density from balbes
look some reasonable.
So i have refined it in ccp4i and phenix differently and then model
build it by coot. My protein has 377 Amino acid. But, I stucked at R
free 41 and now i have amino acid 50 to 369 in both chains, still 49
amino acid at N terminal and 8 amino acid at C terminal are missing
and some loops are missing in between too.
*I have tried the ARP/wARP, but it does not work for it. *
*
*
*I have even tried phase and build of phenix but condition remain
same. i got stuck at R free at 42 and same around 49 amino acids at N
terminal and 8 amino acids at C terminal and some internal loops still
absent.*
*
*
*Thank you in advance.*
*
*