I have never tried a 4D data set in analysis, but it used to be feasible in ANSIG, so we'd better make it possible!
Blocking is undoubtedly important. Justin's advice covers it - if anyone's processing in Topspin on the other hand investigate the xdim parameters.
If your expt is as you say and ends with acquisition on amide Hs, then be sure to chop down the spectrum to just the relevant part (AZARA upper & lower, Topspin strip transform, NMRpipe will have something similar).
Consider storing the computed contours once you've found reasonable levels - that's how ANSIG managed good performance on now ancient hardware.
Dr. Brian O. Smith --------------------------- Brian Smith at glasgow ac uk
Institute of Molecular, Cell and Systems Biology & School of Life Sciences,
College of Medical, Veterinary & Life Sciences,
Joseph Black Building, University of Glasgow, Glasgow G12 8QQ, UK.
Tel: 0141 330 5167/6459/3089
----------------------------------------------------------------------
The University of Glasgow, charity number SC004401
________________________________________
From: CcpNmr software mailing list [[log in to unmask]] on behalf of Wayne Boucher [[log in to unmask]]
Sent: 26 September 2016 17:01
To: [log in to unmask]
Subject: Re: 4D data analysis in CCPN
I’m not sure it will ever cope with such large data files, but if anyone else has any experience with this (Gb sized data files) then please comment.
It sounds like you have a heck of a lot (8k?) points in the acquisition dimension. Possibly you could reduce that to 512 or something (yes, it seems a horrible thing to do). That would still be a big file but not so big.
People can correct me on this, but I guess ft4 is not blocked, so looking at orthogonal planes (so not the fastest two dimensions on disk) is going to be slow no matter what.
Wayne
> On 26 Sep 2016, at 16:58, Wayne Boucher <[log in to unmask]> wrote:
>
> Message from Braden Roth (Jiscmail is being a pain so I’m sending):
>
> To the CCPN community,
>
> Here is my overall question: How to people analyze 4D data in CCPN?
>
> Here is my history:
> I have collected a non-uniformly sampled 4D [13C-15N] HSQC-NOESY-HSQC with 80x80x128 indirect points and processed the data using the NMRPipe plugin, SMILE. The data looks great but very large (my .ft4 file is ~30 GB). I am running 64-bit CCPNMR version 2.4.2 on a 2014 MacBook Pro running El Capitan (see screenshot below)
>
> I can import the spectrum but my project becomes so slow that it’s impossible to use the data in any practical manner. For instance, it takes 10-20 seconds to pick single peak. I have tried to change the graphics handler from Tk to OpenGL but CCPN crashes (as documented in posts dating as recently as 2012). I have also tried using a linux box with Nvidia GeForce (cuda) video card, 6 GB system memory, and an i7 processor. I see no improvement in performance. In the case of my linux system, I’m again running a 64-bit version of Analysis 2.4.2.
>
> My backbone and sidechain assignments have been made in CCPN so i’m hoping there’s a way to continue using that software with my 4D data. Can anyone offer advice for 4D data analysis in CCPN?
>
> Thanks,
> Brady
|