- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Daniel Roe <daniel.r.roe.gmail.com>

Date: Sun, 21 Jul 2013 10:02:32 -0600

Hi,

If you want to calculate a coordinate covariance matrix use either the

'matrix covar' or 'matrix mwcovar' (for mass-weighted) commands. If

you want to calculate a matrix of the correlation between atomic

positions (which is also called 'dynamic cross-correlation map) use

the 'matrix correl' command, which calculates matrix entries as:

Mij = (<ri * rj> - <ri><rj>) / sqrt( (<ri^2> - <ri>^2) * (<rj^2> - <rj>^2) )

See the AmberTools 13 manual for more information about these

commands. Hope this helps,

-Dan

On Sat, Jul 20, 2013 at 8:51 PM, Hongyi WU <why0713.sjtu.edu.cn> wrote:

*> Daniel,
*

*>
*

*> Sorry for my terribly late reply...
*

*> But how can I calculate the correlation between 2 vectors (e.g. 2 alpha
*

*> carbons) over a period of time, which I think is to calculate the
*

*> correlation between 2 matrix:
*

*> {{x1(1),y1(1),z1(1)},{x1(2),y1(2),z1(2)},...,{x1(t),y1(t),z1(t)}}
*

*> And
*

*> {{x2(1),y2(1),z2(1)},{x2(2),y2(2),z2(2)},...,{x2(t),y2(t),z2(t)}}
*

*>
*

*>
*

*>
*

*> Hongyi
*

*>
*

*>
*

*>
*

*> -----ÓÊ¼þÔ¼þ-----
*

*> ·¢¼þÈË: Daniel Roe [mailto:daniel.r.roe.gmail.com]
*

*> ·¢ËÍÊ±¼ä: 2013Äê7ÔÂ9ÈÕ 22:08
*

*> ÊÕ¼þÈË: AMBER Mailing List
*

*> Ö÷Ìâ: Re: [AMBER] ´ð¸´: timecorr
*

*>
*

*> Hi,
*

*>
*

*> The poorly-named 'crosscorr' command in CPPTRAJ will calculate the Pearson
*

*> product moment correlation coefficients between 2 or more data sets. For
*

*> example, using the input:
*

*>
*

*> parm ../tz2.parm7
*

*> trajin ../tz2.nc
*

*> distance d1 :2 :12
*

*> radgyr rg :1-3 mass nomax
*

*> crosscorr d1 rg out d1rg.dat
*

*>
*

*> You'll get a 2D matrix as output:
*

*>
*

*> #crosscorr_00002
*

*> 1.000 1.000 0.0000
*

*> 1.000 2.000 -0.3105
*

*> 2.000 1.000 -0.3105
*

*> 2.000 2.000 0.0000
*

*>
*

*> Here the first two numbers are the indices of the data sets (corresponding
*

*> to the order you gave to the crosscorr command), so the first line reads
*

*> 'correlation coefficient between data set 1 (d1) and data set 1 is 0.0 (of
*

*> course it should actually be 1.0, but since we know that it's not actually
*

*> calculated which is why it's reported as zero), the second is correlation
*

*> between data set 1 (d1) and data set
*

*> 2 (rg), etc.
*

*>
*

*> Note that the Pearson coefficient is also reported as part of output from
*

*> the 'corr' command (which computes time correlation functions for non-vector
*

*> data sets), e.g.:
*

*>
*

*> 1: [corr d1 rg out cross.dat ]
*

*> CORR: 101 elements, max lag 101
*

*> CORRELATION COEFFICIENT d1 to rg IS -0.3105
*

*>
*

*> Hope this helps,
*

*>
*

*> -Dan
*

*>
*

*> On Tue, Jul 9, 2013 at 6:34 AM, Hongyi WU <why0713.sjtu.edu.cn> wrote:
*

*>> Dear David,
*

*>>
*

*>> Thanks for your answer.
*

*>> But I get the definition of correlation from our text book, Molecular
*

*>> Modeling Principles and Applications by Andrew R. Leach. And it's just
*

*>> the same as Pearson correlation (or at least the mathematical form). I
*

*>> have attached a screenshot of the definition to this letter.
*

*>>
*

*>>
*

*>> -----ÓÊ¼þÔ¼þ-----
*

*>> ·¢¼þÈË: David A Case [mailto:case.biomaps.rutgers.edu]
*

*>> ·¢ËÍÊ±¼ä: 2013Äê7ÔÂ9ÈÕ 20:03
*

*>> ÊÕ¼þÈË: AMBER Mailing List
*

*>> Ö÷Ìâ: Re: [AMBER] timecorr
*

*>>
*

*>> On Tue, Jul 09, 2013, Hongyi WU wrote:
*

*>>>
*

*>>> The function timecorr in AmberTools 13 will return squared
*

*>>> correlation r^2, right?
*

*>>
*

*>> No: the "time correlation" [C(t)] is unreleated to the Pearson
*

*>> (linear) correlation function (which it sounds like you might be thinking
*

*> of).
*

*>> C(t) is normalized such that C(0)=1; it is not the square of anything.
*

*>>
*

*>> Apologies if I am misunderstanding your question.
*

*>>
*

*>> ..dac
*

*>>
*

*>>
*

*>> _______________________________________________
*

*>> 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
*

*>>
*

*>
*

*>
*

*>
*

*> --
*

*> -------------------------
*

*> Daniel R. Roe, PhD
*

*> Department of Medicinal Chemistry
*

*> University of Utah
*

*> 30 South 2000 East, Room 201
*

*> Salt Lake City, UT 84112-5820
*

*> http://home.chpc.utah.edu/~cheatham/
*

*> (801) 587-9652
*

*> (801) 585-9119 (Fax)
*

*>
*

*> _______________________________________________
*

*> 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
*

Date: Sun, 21 Jul 2013 10:02:32 -0600

Hi,

If you want to calculate a coordinate covariance matrix use either the

'matrix covar' or 'matrix mwcovar' (for mass-weighted) commands. If

you want to calculate a matrix of the correlation between atomic

positions (which is also called 'dynamic cross-correlation map) use

the 'matrix correl' command, which calculates matrix entries as:

Mij = (<ri * rj> - <ri><rj>) / sqrt( (<ri^2> - <ri>^2) * (<rj^2> - <rj>^2) )

See the AmberTools 13 manual for more information about these

commands. Hope this helps,

-Dan

On Sat, Jul 20, 2013 at 8:51 PM, Hongyi WU <why0713.sjtu.edu.cn> wrote:

-- ------------------------- Daniel R. Roe, PhD Department of Medicinal Chemistry University of Utah 30 South 2000 East, Room 201 Salt Lake City, UT 84112-5820 http://home.chpc.utah.edu/~cheatham/ (801) 587-9652 (801) 585-9119 (Fax) _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Sun Jul 21 2013 - 09:30:02 PDT

Custom Search