Hi Hiroshi,
I am processing time-course data right now (but this is applicable for all studies). I am finding a number of features with poor spectral matching in the alignment result, but when I check the "Spectrum reference file name" column I find the spectra are commonly from samples with very low levels of the metabolite. If I then go to a file with high levels of that same feature I find a very highly scored match.
So my question is: Is there a way to force MS-DIAL to select the MS2 spectra (in the aligned peaks list) from the sample with the most intense response?
Perhaps a relevant note: This feature found using an mz-RT list.
Thanks,
Brian
I did find the 'relative abundance cut off' under advanced on the identification tab. I think this will solve the problem.
Unfortunately the 'relative abundance cut off' appears to actually affect the abundance cutoff of the MS2 spectra and does not force samples with more abundant peaks to be chosen for similarity matching.
It appears the original problem listed above does persist, any advice on the matter would be greatly appreciated.
I have the same confusion as well.
Hi Brian,
this issue is bit actually strange for me as well...
Can you send me a small-scale data set to check this issue?
Even though I checked my dataset, I could not find such a similar issue.
Hiroshi
Hi Hiroshi,
Thank you for your quick reply. I have sent a google drive link to the Riken email I have for you. It is from my CZ Biohub email.
I believe I have included all files relevant for you to open the study yourself. However, if you think it will be faster and easier to assess via teamviewer and zoom I am happy to accommodate the time difference by staying up late or getting up early.
Alignment ID 2004 is the example I have illustrated above, but there are a number of occurrences I have observed including alignment IDs: 1401, 4083, 4688, 1159, 2221, 5111, 1902... I can find more examples if needed.
I can confirm my note from the first post, this issue has only been observed in features identified by mz-RT list. Features solely identified by the .msp file select spectra from abundant samples as far as I can tell.
Thank you so much for you attention on this matter!
All the best,
Brian
Hi Brian:
your note: >> this issue has only been observed in features identified by mz-RT list
was very helpful for me.
This is the reason why you see the current issue, and actually, I cannot resolve the issue at this moment.
As you may know, the representative file for each alignment spot will be determined by the identification score followed by the ion intensity among samples.
In this case, when you use a text library, sometimes (or often),
a peak feature having vey low intensity may obtain the highest identification score because of the good consistency of its retention time and m/z although the intensity and its MS/MS spectrum are very low.
Although I cannot resolve this issue at this moment because I will release completely new msdial program within this year, I will resolve this in the next version of ms-dial program!
Hiroshi