Does the facility that ran your samples have any training support available for data analysis? This might be helpful for getting you started. You can also try to follow the MS-DIAL tutorial if you haven't already: https://mtbinfo-team.github.io/mtbinfo.github.io/MS-DIAL/tutorial.html
To try to answer your questions:
1. The most important thing is that you choose a library that matches the way your data were acquired. Were the data acquired with GC-MS or LC-MS/MS? If the latter, was it in positive or negative mode? If, for example, your samples were analysed using LCMS/MS, with positive mode electrospray ionisation (ESI), you could use the "ESI(+)-MS/MS from authentic standards (16,481 unique compounds)" library from MSDIAL's set of curated libraries. The facility might also have access to a more comprehensive commercial library. If you are unsure how your samples were analysed, you should check with the person who acquired your data.
2. The libraries we are talking about above are what is required in the identification tab - if you download from MSDIAL it will be in .msp format, which is basically just a text file. But if you have an excel file with retention times from your samples, it sounds like the facility has already done some data analysis of your samples for you? If so, depending on your needs you might not need to use MSDIAL at all. Hard to say with info in your post.
3. Yes that is OK, MSDIAL processing will run without standards, or QC samples for that matter. Standards are useful if you want to confirm the identification of particular metabolites or quantify the metabolites in your sample, but it is probably OK not to have them for a qualitative, untargeted metabolomics project.
I have heard that version 4.24 works for some people with these issues, but it doesn't fix the problems for me.
I don't do lipidomics, but I've found the most reliable way for me to avoid crashes is to first run alignment steps without any identification, and then go back and run identification after saving the alignment file. For some reason it seems to crash less frequently this way for me.
I don't know of a way to do this within MS-DIAL, so I would untick the 'use retention information for scoring' box in the Identification tab of the parameter settings, and then compare the calculated retention indices with your putatively identified compound RIs manually.
If there are so many compounds that it isn't feasible or efficient to do it manually, the webchem R package has functions for searching online databases for RI information of the compounds in your dataset and allows you to specify column polarities and retention index types - meaning you could specify only polar Kovats retention indices and then quickly screen for big differences.
What libraries are you loading on MS-DIAL for your LC-MS data. For GC there seems to be a lot of help in compound annotation but not as much for LC, I feel. Wondered if your approach may help me use MS-DIAL for LC-MS data too.
Sorry for the delay. I just used the .msp libraries available from the MS-DIAL database page, which has a bunch of freely available spectral libraries for both positive and negative ionisation. I haven't spent much time using MS-DIAL for LC-MS though, and from what I did do, most of the compounds I was interested in were not correctly annotated. I didn't really try to get it working well though, so this may well be partly error on my part.
I don't have a good solution, but wanted to say that I also have this issue for GC-MS data. It seems to crash at some point before showing the peak spot viewer or creating any .mtd2 file about 9 times out of 10 - but every so often, after stubbornly fiddling with the parameters and re-attempting, the alignment works and the peak spot viewer pops up.
I have not figured out which parameters, if any, are actually important for getting it to work. Several times I have thought I discovered the secret, only for the next analysis to crash again - and again and again.
I have accepted the stochasticity of getting MSDial to work and figure it is better to indulge my gambling habit here than at the slot machines, but would certainly appreciate a solution if you or anyone else has figured it out.
I have a similar issue, but for me it crashes before ever showing the peak spot viewer.
This only seems to happen when using GC-MS data, and only when loading 10 or more files. I can load and process up to 9 files successfully, but it reliably crashes when loading 10 or more. It doesn't seem to matter what parameters are used or whether any libraries are searched, etc. This is the case with either .mzML or .D files.
I can load at least 20 LC-MSMS files though (and presumably more, but I haven't checked), so maybe the problem is specific to GC-MS processing? What type of data are you using?
