Skip to main content
Topic: molecular networking (Read 112 times) previous topic - next topic

molecular networking

Hello Hiroshi,

I have some questions regarding the generation of molecular networks within MS-DIAL, as -->Export --> Molecular spectrum networking export:

1) does the similarity cut-off represent the edge cosine score (exactly as in GNPS)?
2) is it possible to set a minimum of matched peaks in the spectra? e.g. 4-5 peaks, and exclude consensus spectra with less than a threshold?
3) could you add more explanations relevant to these settings and different cut-offs (in their respective fields, as "pop-up" tips in the menu, but also if possible in the MS-DIAL tutorial page)?

Thank you so much

Stefano

Re: molecular networking

Reply #1
Hi Stefano:

1) does the similarity cut-off represent the edge cosine score (exactly as in GNPS)?
Probably, no. At least, the concept is same, but the algorithm detail should be different from GNPS.
See the method detail in Nature Methods 2019.
https://www.nature.com/articles/s41592-019-0358-2

2) is it possible to set a minimum of matched peaks in the spectra? e.g. 4-5 peaks, and exclude consensus spectra with less than a threshold?
OK, I will make it in near future.

3) could you add more explanations relevant to these settings and different cut-offs (in their respective fields, as "pop-up" tips in the menu, but also if possible in the MS-DIAL tutorial page)?
Is this not enough yet?
https://mtbinfo-team.github.io/mtbinfo.github.io/MS-DIAL/tutorial#section-6-3

Hiroshi



Re: molecular networking

Reply #2
Hi Hiroshi,

Thanks for considering all these suggestions.

About:
3) could you add more explanations relevant to these settings and different cut-offs (in their respective fields, as "pop-up" tips in the menu, but also if possible in the MS-DIAL tutorial page)?
Is this not enough yet?
https://mtbinfo-team.github.io/mtbinfo.github.io/MS-DIAL/tutorial#section-6-3


I am familiar with GNPS and FBMN, still I am not completely sure if I interpret correctly (and how to adjust accordingly) these 3 different parameters:

1) - Relative abundance cut off [%]. e.g. 1  --> is this the threshold for the ion intensity, i.e. the higher the set value the fewer the spectra kept into consideration in building the network? What would you suggest as a default value?
 
2) - Similarity cut off [%] --> this is clear (i.e. similar to the cosine score in GNPS)

Check-box (Ticked) -Export ion abundance correlation among samples
3) - Similarity cut off [%] --> does this refer only to the "check-box" above, for exporting ion abundances correlations?

I think some  "pop-up" tips like your nicely put for other data-processing fields it would help also other users who perhaps never run a molecular networking before.

Thank you once again!

Stefano