If you imagine a pixel for each comparison 260K x 260K pixels is pretty high resolution... Usually 2DRMS is to try to get an understanding of the conformational overlap and you really do not need to see each frame to each frame.
My approach would be to look at every 100th or 1000th frame. If concerned you are "missing information", do it twice. If every 1000th frame, do frame 1->260K by 1000 and compared to 500->260K by 1000. In trajectory loading you can set the offset.
This same adage holds for clustering in ptraj. If you try to cluster over all frames at once there are simply too many pairwise comparisons. So, what we normally do is "sieve" or select every 1000th frame for initial cluster and then go back over the data to add missing frames to existing clusters. You can sieve at regular offset or randomly. As far as how many frames to skip over, depends on memory and how long you want to wait (i.e. every 10th frame, every 100th, ...).
In the old days, we write Postscript files directly and when we generated the big ones (too many frames being compared) it took forever to visualize or print and did not provide any new information.
--tec3
________________________________________
From: Saikat Pal [saikatpaliitg.yahoo.com]
Sent: Friday, September 21, 2018 6:03 AM
To: AMBER Mailing List
Subject: [AMBER] Query on 2DRMSD
Dear all,
I want to perform 2DRMSD for 260000 frames. But the problem is that it takes a huge memory and also crash the run. Is there any solution ??
Thanks and Regards,
Saikat Pal
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Sep 21 2018 - 17:30:02 PDT
