Dear Dhanunjaya Mitta,

The PRIDE Archive is the best proteomics data repository.

https://www.ebi.ac.uk/pride/archive/

Best regards,

Leite

Thanks @Giuseppe G.F.Leite

Hello @ Giuseppe G.F.Leite thanks for your reply. But what my question was i would like to know the type of data we will get from the Mass Spectrometry and how can we use those data in the Machine Learning Algorithm???

The data are an abundance of signals recorded for specific mass to charge ratio. This allows to identyfy compounds on the basis of spectral libraries.

Machine learning algorithm can be defined to find correlation between observed characteristics of a mixture of compounds analysed with their mass spectra. If you use GC-MS this is "two-dimensinal" separation a one by GC (mixture to separate compounds) and second one by MS (ions produced from each compound are recorded separately). For example you can correlate a odorous character of some types of mixtures of volatiles like wine aroma with their composition determined by GC-MS. If you do not use GC only MS, so the correlaton will be based only on some ions. At end on the basis of MS signals you predict the aroma.

regards,

G

See attached

See attached

gcms gives several pieces of data. GC is a chromatographic technique which gives a retention time and peak shape. Comparison with these two data with that given by a known standard run under the same conditions at a time near to the analysis gives a suggestion as to identity. Peak shape can mean the width at half the height of the peak, if tailing occurs, if it is symmetric, etc. The second data is the mass spectrum which is a plot of fragment masses and the intensities of these masses. Taking the ratio of fragment intensities of each fragment mass to the intensity of the strongest fragment can be compared to the same ratios obtained from the known standard used above. There are tables as to what are considered "good" matches. Also, the presence or absence of fragments in the unknown relative to those in the standard mass spectrum has importance.

Futhermore, when using library standards, a quality of match is provided. In forensic science, usually you need at least 90 percent match for a good match, 95 + percent match is very good. But, in forensics, it is still considered indicative and should be compared with the standard run at the same time for positive identification if such standard exists. In other fields, library matches may be more acceptable. Optical isomers are not distinguished unless special techniques such as derivitization or chiral columns are used so still may have cases where final answer is not given by gcms.

Thanks @cyril ogbuanu and @wayne morris.

@cyril ogbuanu: How can we know the molecular compound by using a given mz value and intensity values... Is there any tool to calculate it automatically??? Or do we need to do it manually

Dhanunjaya,

generally the highest m/z on the spectrum is the molecular ion. The electron impact ionisation is a "hard ionization" thus besides the molecular ion also the fragment ions are present. This phenomena is reproducible, that's why you can compare it with library which matches the spectrum and gives a prediction of possible compounds.

Sometimes the molecular ions are not abundant enough.

in this case a chemical ionisation can be used, but the identyf. is more problematic. In this case the molecular ions are often much more abundant but the library data are not that big like for EI (6-10 mln. of compounds).

regards,

G

Dear Dhanunjaya Mitta,

The main data we can get from the MS are chromatogram for the molecular weight. We can use the molecular weight to speculate the molecular structure and formula. If need the further confirmation, we can use the standard compound and MS/MS (or LC-MS/MS). All above procedure will generated the data and represent by chromatogram and peak area (signal response).

regards,

R