Re: [sdpd] Automatic peak position hunting

[wilsond212...@aol.com]

In a message dated 12/21/2000 6:10:11 AM 
Eastern Standard Time, 
armel...@fluo.univ-lemans.fr writes: 


> Hi, 
>
> It seems that automatic peak position hunting programs 
> like PowderX, EVA 5.1, Winplotr... are unable to 
> distinguish two 
> adjacent peaks when the maximum of 
> these 2 peaks are separated by only one 
> step, even if 
> the 2 peaks are clearly separated by eyes, seeing à 
> M-like 
> shape at the peak top like this : 
>
>                             |\/| 
>                             |  | 
>           _________|  |__________ 
>
> Comments ? 


I have no experience with any of 
these programs, but ran into similar 
problems with commercial programs for 
plotting and interpreting HPLC 
chromatographs some years ago.  There are 
several possible problems: 
1. The program may be smoothing the digital input 
for its interpretation by 
combining consecutive points. 
2. The peak 
detection may be set to require too many consecutive points to be 
below the 
maximum before a peak is detected. 
3. The signal strength at peak onset may 
be part of the calculations to 
locate the peak, e.g., peak location is 
midway between onset and end-of-peak, 
and the peak end requires going back 
to the signal strength at peak onset. 
4. The peak may be determined by when 
zero slope is detected, and the slope 
calculation extends over more than two 
adjacent points. 
5. Was the baseline noisy?  The program may be 
automatically determining the 
standard deviation of the signal from the 
baseline noise, and the minimum 
between the two peaks may be within that 
statistical range of the maximums. 

In any event, the problem could only 
be answered in my case by a very 
detailed examination of the peak-finding 
code.  If the programs you mention 
have open code, that might be possible to 
do.  If the code is proprietary, 
I'm not sure what the answer is. 

Wilson DeCamp