Good afternoon! I have run a calibration curve and each levels were injected 12x for replication in Agilent 6545 QTOF All ions mode. I used Masshunter Qual to identify some molecules from our standard mix and then compared this to MS-Dial. We were hoping to use MS-Dial with our future untargeted data analysis but this is just proof of concept on checking data quality between MS-Dial and Masshunter Qual/Quant from Agilent.
Sample info:
Calibration curve in methanol: 0, 0.05, 0.1, 0.5, 1, and 2 ppm with 12 injections each levels and ran this in All ions pos and neg in Agilent 6545 HILIC chromatography.
Problem: I am seeing high %CV's from MS-Dial analyzed data than in Masshunter Qual. I was hoping MS-Dial would give the same if not better quality than Masshunter Qual. Can some one please check my parameters?
For example, at 2 ppm for Arginine neg mode, we are seeing 8.4% CV while in MS-Dial we were seeing 90% CV!!! The retention times were comparable between the two data analysis tools. What is causing MS-Dial to have such high %CV? Your help would very helpful. Arginine and Methionine calibration curves are shown in the attached files. Thank you and looking forward to your response. Keep safe!
how do your extracted ion chromatograms look for your compounds of interest in MS-DIAL? maybe you need some smoothing? are you picking up correct feature for your calibration curves?
Hello! Thank you for the reply. I ran my samples in Agilent 6545 and so I used their tool to extract the ion chromatograms. And so I know that the peaks are there consistently but MS-Dial is not picking them up consistently. Hence, I am getting noisy data. I have sent my MS-Dial parameters in the attachment. Can someone please check them? Thanks!
can you share a raw data file (google drive, dropbox etc.) to reproduce the problem with your MS-DIAL parameters?
Hello! Ok. CAn you please send me your email address so I can put in the dropbox and send it to you? Thank you!
HI Romanas! Good morning. =) I just sent you 6 files of negative mode all ions data. These are separate injections of 2 mg/L standard concentration from different vials. Please keep me posted. Thank you and keep safe.
I think the problem is the SCAN order (e.g. use mzML to check in SeeMS from ProteoWizard or MZmine). We are also using Agilent QTOF - the MS files start with AIF scans. Try the experiment settings as in attached screenshot.
Good morning, Romanas! Thank you. I will try that. But I am seeing that you are not getting reproducible EIC for phenylalanine either using MS-Dial even with the new experiment setting. Can you please verify? Should I send a non-converted files? I appreciate your help with this. Thank you, Romanas!
Please see attached picture of the EIC of phenylalanine from masshunter.
Hi Romanas,
I did what you have asked me and I am seeing the same irreproducibility. Please check my attachment. Maybe I should check the converter. Thanks!
Marites
Hi Marites, can you please send the original .d files? The height/area might be slightly different due to smoothing parameters/integration in different softwares. What do you exactly mean by "irreproducibility"?
Hi Romanas! Good afternoon! Sorry for the delayed response. I am just coming back my vacation. I am going to send you that data asap. By the way, do you think we should turn off the retention time tolerance since our gradient may differ from the library? Thank you! Hope you've been well and keep safe.
Hi Romanas,
Thank you for letting me know that you got my 3 technical replicate data on 2 ppm negative mode HILIC assay I have attached some of the data processed from Masshunter (molecules' area, height, RT and their corresponding %CV's, which are very low). Scaffold and XCMS have similar low %CV's while MS Dial has %CVs >30%. I have doubled checked the parameters and tweaked them but %CV's were still high. I hope you can help figure out what's going. My parameters used are on previous attachments. Thank you and keep safe.
Regards,
Marites
Hi Marites,
I think I found the problem - it is in the number of spectra in each experiment. Sometime around 0.25 min there are some scans "missing" and the number of missing scans seems to be different for each file (see attachment). However, MS-DIAL assumes that the number and order of scans is consistent and hence all the problems with intensities/CVs...
@Hiroshi Tsugawa - Probably the readin/assignment of collision energy has to be modified or at least a warning given if the number of spectra is varying between to MS1 spectra.
@7clouds7 - what is happening around 0.25min in your method?
Best,
romas
This is awesome, Romanas! For 0-0.2 mins, LC stream is going to waste to protect the instrument from any salt contamination. Then, at 0.2 min LC stream switches back to MS. Please let me know's the remedy for this. Thank you so much, Romanas!
probably the having multiple time segments in the method causes the issue. For the future try acquiring all data in one segment, or stop acquiring data in the 0-0.2min segment. Anyway, are you sure your flow-through peak with salts comes around 0-0.2min?
for the current files, if you want to use the MS2Dec and CorrDec in MS-DIAL, I am afraid you will have to check and crop (subset function in ProteoWizard MSconvert) the files one by one to make sure they match MS-DIALs experiment settings file. However, if you just want MS1 then proceed with simply filtering MS1 - no deconvoluted MS2 spectra in such case.
Thanks, Romanas! And so you recommend for us to use proteowizard to convert our files and not use ABF converter?
Also, we usually do not acquire anything from 0-0.2. It just goes to waste. And so, just to be clear, currently, there is no remedy yet for this. I am glad we found the culprit but too bad there is no remedy yet. What is the process for requesting this for MSDial? Thanks and keep safe.
HI Romanas,
I tried using proteowizard and just only focusing on data from 50-500 seconds. But unfortunately, I am still seeing the high %CV's. This is a bummer. I have included the picture of the proteowizard parameters and also some of the amino acids that normally had low %CV's (phenylalanine and glutamic acid) using Masshunter but I am still seeing high variation in MSDial. But when I did just regular MS1 without any deconvolution, I am getting really good %CV's.
Hi Marites, instead filtering by seconds in ProteoWizard, filter by scans. The key point is to set the beginning of the filtering range so that output file matches the MS-DIAL settings.
Thanks, Romanas! Can you please guide me as to how to filter by scans instead of by seconds? I have not done this before. How would I get the # of scans at 0.25 minutes? I have the acquisition rate which is 4 spectra/seconds, which means there are 60 spectra at 0.25 minutes. Should I then filter from 80 scans - 2000 scans? I chose 80 just to make sure I am out of the inconsistent range. Thanks!