I would like to report some issues with respect to the export of normalization results.
Following data processing with MS-Dial (4.80) I performed LOWESS normalization. For the data set measured in negative polarity everything works fine, and I was able to export the normalization PDF without problems. And the plots in the PDF were okay. For the positive data set I used the same procedure but finally a problem occurred during plotting/export. The process needs a lot of time. At the end it crashed, and MS-Dial was closed immediately. I repeated this several times but without success.
I wonder what the problem might be, because for negative dataset it worked. The size of both datasets and the respective file sequences were equal. The only difference was the number of ´Ref. matched´ features (pos: 2867 vs. neg: 839). The PC I used for data processing was also quite powerful (iCore i9, 12 cores, 128 RAM, Win10, 64 bit, engl.), and the utilized working memory was approx. 20% as the process aborted.
As we planned to use the normalization function for future studies, it would be nice, if some someone could help me with this issue.
I have a question regarding the normalization plots, that can be exported from MS-Dial, in order to assess normalization procedure (at the moment, I used LOWESS).
As I understood right, these plots are generated only for ´Ref. matched´ features and TOP100 (Int.) of ´Unknowns´. But, in my opinion it would be interesting also to have plots of TOP500 (´Unknowns´) and of selected (statistical significant) features.
Is there currently a way to get access to additional/individual normalization plots in MS-Dial?
MS-Dial version: 4.80 I have a question regarding the PCA function. We have processed our LC/MS data set, but following data processing we can not select the PCA function (greyed out). This was independent whether we check the gap filling function. We also have filled in the class IDs prior to processing (in total, we have 8 groups: 1 x Blank group, 1 x QC group, and 6 x Sample groups)
Why this is the case, and what I have to do for being able to use the PCA function?
two additional questions for my understanding: 1) Data processing in MS-DIAL is done the following way, isn´t it? a. peak detection b. identification (annotation of the peaks in all samples if possible) c. alignment (und selection of top annoation hit by means of the highest total score)
2.) if you re-process your data starting from "identification" (e.g. by using another MSP-File) which of the following statements is true? a. The identification step and the subsequent alignment step is re-processed (but not the peak detection step). or b. Only the annotation hit in der respective aligned feature list is modified (no separate alignment step is performed)?
Would you for example suggest the following: if you have 20 or more data points per peak -> use smoothing level 4, whereas if you have less than 10 data points better use smoothing level 2 (or similar)? Just as an idea.
Of course, you have to proof this by your own data.
In that context, one additonal question came to my mind.
Is one single MSMS match across all samples sufficient to get a final identification result for a specfitic feature or do you need several matches (how much)?
Hello, I checked the parameter export and import function (version: 4.80) und would like to report some currently occurring errors.
Parameter export: Following parameter settings were not correctly exported (via “Export > Alignment result”) in the txt-File: - The setting for "Consider Cl and Br elements" is missing regardless of whether you checked this field or not. - With respect to the blank filter option, independently what your settings were, you always get the following output (default setting): "Sample max / blank average 5 Sample average / blank average 5" - Regarding the annotation filter criteria: In the text file, you have a setting for "Keep identified and annotated metabolites", whereas in the GUI you can make selections for "Keep ´reference matched´ metabolite features" and "Keep ´suggested (w/o MS2)´ metabolite features".
Parameter import: Following parameter settings were not correctly loaded (via “Load parameter”) in the GUI: - in the peak detection tab: the smoothing method is often not correctly imported, e.g. although I selected “Savitzky-Golay filter” the “linear weighting method” (default settings) was imported. - in the alignment tab: Instead of my individual settings the default settings of MS1 and RT factor were imported. The “Reference file” is not correct after parameter import (I suggest, here always the first file of the analysis file list is selected). - with respect to the blank filter option you have the same issue as mentioned above probably because the parameter export was not sufficient.
Maybe this bugs can be eliminated in one of the next updates.
I have some question with respect to specific alignment parameters.
1.) My first question refers to the parameters "MS1 factor" and "RT factor". As I understood right, the parameters determine the weighting of mass and retention time (deviations) used for calculating the total score of a feature matching the post-identification process. Up to now, my inetention was that the total sum of both parameters should be 1, e.g. 0.5 for MS1 and 0.5 for RT. However, I recognized that you can process your data also with parameter values whose sum is bigger than 1. So, I am not sure how this weighting finally works?
2.) The second question concerns the “N% detected in at least one group” parameter. Does “one group” also includes the QC and the blank group, respectively? If the value is smaller than the given parameter setting (valid for all sample groups, but not for QC/blank group), what will be the consequence? Is such a feature completely eliminated from the alignment output or is it only eliminated from the sample groups but not from the QC or blank group (and thus it would remain in the feature list)?
I was wondering how to deal with the different setting options for the MS1 tolerance (in Da) for peak detection vs. alignment. Is it generally recommended to have a "looser" setting for peak detection and a correspondingly "stricter" setting for alignment, or should both parameters be set very similarly or even identically?
Moreover, I have a question regarding the function of the BPC mass slice parameter. Does it determines whether data points are within the same slice and then assigned to the same peak? We work with a high-resolution MS (QTOF). Values like 0.1 Da or 0.05 Da as recommended by MS-Dial seem quite generous at first glance. But I may not have fully understood the meaning of this parameter.
Hi, I have a question about the smoothing parameters. According to an older MS-Dial version the recommendation was 1-3, However, for the current version (4.80) it is 2-4. Does this recommendation apply regardless of the selected smoothing method? Could it be simplified that for peaks with >10DP smoothing level 3 to 4 should be the better setting, and only for peaks with <10DP a smoothing level of 2 would be preferable?
I have a question regarding MS2 spectral matching in MS-Dial.
Sometimes, you will have different matched compounds for one and the same feature. Let´s say, when you go back to the chromatographic peak lists you have 4 x matches of compound I (total scores: 90, 86, 86, 80) in samples 1-4, 2 x matches of compound II (Scores: 62, 68) in samples 8 & 11 and 1 x match of compound III (Score: 92) in sample 5.
Which of those matches would then be reported in the aligned output (feature list)? From my feeling, I would suggest compound III, as it shows the highest total score overall. Otherwise, it could be compound II, because it was matched more frequently than compound III. So my major question is, what decision criteria does the software use in such a case?
for identification purpose I used post identification (txt file) as well as MS2 spectra annotation using the open-source msp-file from the website. If I had a closer look into the data, I recognized that in any case when I got a match by post identification there was no respective MS2 match, although most of these features had a good MS2 spectrum. Therefore, and because I know that some of these analytes are present in the spectral library, I expected an MS2 match, too. But only the post identification result is reported in the annotation table and the feature list, respectively.
Can I find the result from MS2 matching of the "postident matching features" somewhere else? If not, what is happening with features that are matched by post identification? Aren´t they considered for subsequent MS2 annotation?
after processing the current data set I found out that it can make a difference for alignment, if you choose an msp file in the identificaton step vs. if you don´t.
I tested this for the current dataset in positive mode. By selecting an msp-file (the open-source file of MS-Dial) I got 14464 features. When no msp file was selected, I only got 14106 features after processing, although the same parameters were used. After all the difference in features is about 2%.
At the moment, I can not explain, how this can happen because the alignment should work the same whether you select an msp-file or not, isn´t it?
