AMDIS: Disqualifying library hits based on retention time

[sonicboom12345]

Hi all,

I'm sad to report I'm having a bit of trouble with AMDIS. I have a library of compounds I'd like it to search, but some of them are spectrally similar, and as a result, the software is mis-reporting. To solve this, I've incorporated retention time information into each library entry:





And I've instructed the software to qualify hits based on retention time, with a 0.2 min RT window, and a 100 point penalty to any and all compounds falling outside of that window:





...but no matter what I do, the software insists on applying (at most) a -12.3 correction based on retention time. Take note of the following image, where a peak eluting at 10.13 minutes is being identified as the library compound with a retention time of 9.44 minutes. This is well outside the 0.2 minute RT window, yet the compound is not being penalized to the extent I would like it to be.





I need to take this library and these settings and use them to run a batch job on ~500 chromatograms. The results will then be subject to chemometric statistical analysis, so it's very important that problems like this not be allowed to crop up, or else the whole entire study might be thrown off. Can anyone please advise me on how to correct this? I would like it to immediately disqualify any library hit that falls outside of 0.2 minutes from the expected value.

Thanks!

[chemstation]

Hi,
I suggest you switch to using retention indices, as the method to discriminate. It has more variables and is known to work (I use it all the time).
As using a RT value and expecting them to not change over 500 runs, particularly if instrument maintenance has been performed, is hopeful.
I am presuming that some sort of QC was run before each sequence, so you can monitor any changes in the system.
Kind regards
Alex

[sonicboom12345]

Hi,
I suggest you switch to using retention indices, as the method to discriminate. It has more variables and is known to work (I use it all the time).
As using a RT value and expecting them to not change over 500 runs, particularly if instrument maintenance has been performed, is hopeful.
I am presuming that some sort of QC was run before each sequence, so you can monitor any changes in the system.
Kind regards
Alex

Thank you kindly for your reply, Alex.

We've already done the runs, and the retention time does not drift significantly over the course of our data set. At the very most, our peaks are displaced by ~0.02 minutes. (Internal standard elutes at 4.11 minutes in the earliest runs, and 4.13 minutes win the latest runs.)

Although retention indices are certainly superior to retention times, I don't know that utilizing them would improve our results all that much in this case. Having to calculate and input retention index values into our rather sizable AMDIS library is also something I would not relish having to do.

I appreciate your good advice. Do you have any insight into why the settings described in the original post might be failing? I will look into retention indices if it's absolutely necessary, but I would prefer to make it work using retention times if it's at all possible.

[chemstation]

I had a look at my AMDIS v2.72 running under WinXP, and when I created a test library, and I changed the RT to emulate your values,
I could get AMDIS to behave as expected using the ‘Use retention time’.
When I then entered in the RI data, I could control it as expected when using ‘Use RI calr.Data +Internal Std.
When I then switched the settings back to using ‘Use retention time’, the correction data stopped working properly ,
and hit a maximum of -5.4, regardless of what settings were entered.

When I looked at my AMDIS v2.70 running under Win7, I saw something similar except (maximum of -10.3),that if I changed the RT window from 0.2 to 0.1,
the correction value would increase (maximum of -25.3).
So there is definitely a bug in AMDIS.

I still recommend that going to RI will solve the issue.
IMHO, my suggestion would be to create an excel spread sheet, that when you copy the RT from the library,
and paste into excel, the RI would automatically be calculated, that you can then copy and paste back into the library. (no transcription errors)

[sonicboom12345]

Hello again,

I am looking into utilizing retention indices (as opposed to retention times) to conduct my analysis. However, I have never done this before using AMDIS, and I have found the documentation to be somewhat lacking. I would greatly appreciate if you could offer some guidance as to how to set this up (what settings to use, which buttons to press).

We do not do a QC run before each sequence, rather we include a single internal standard in each sample so we can monitor instrument performance based on the elution time, peak shape, and integrated peak area of the I.S. I am wondering if this can be used to calculate the retention index, as I see AMDIS has an option to "Use Internal Std for RI." What is the process here? Do I simply create a second library with the mass spectrum of my internal standard as the sole hit, and direct the software to utilize it as my internal standard library?

Even then I am unclear what mathematical formula AMDIS uses to calculate retention indices off of it. You suggested that I create an Excel spreadsheet to help automate the conversion of retention times to retention indices -- but what precisely operation should I carry out to do this? I would be grateful if you could paste the formula, it would be very helpful to a neophyte such as myself.

Thank you.

[chemstation]

Hi,
You need to run a series of n-Alkanes, using the exact same method as that used to acquire the data.
you cannot use a single data point, as AMDIS calculates RI based upon saved RI information.

04070 SIGMA-ALDRICH
Alkane standard solution C8-C20

04071 SIGMA-ALDRICH
Alkane standard solution C21-C40

These will give you the retention times that you need for AMDIS and Excel,

The formula for RI calculations is within the AMDIS help file,
if RTC8 is the retention time for C8, and RTC9 is the retention time for C9, and RTUK, is for the UnKnown that came out
between C8 and C9
then RI = ( (RTUK-RTC8) * ((RTC9 - RTC8)/100)) + 800) or something, just going off memory.

regards
Alex