I must sadly and repeatedly say that HPLC-MS is not a quantitative method at all.

Binding assay of RIA or ELISA using IgG and assay using Avidin/Streptavidin are also not the quantitative methods at all.

Absorption of molecules to other materials has occurred mainly by the van der Waals force. To achieve the high recovery analysis resisting to the strong van der Waals force is the essential point in the quantitative analysis. Therefore, I have previously performed the recovery test in ODS-column separation or RP-HPLC (please see file; Lysozyme by RP-HPLC). Both HPLC-MS and ELISA methods have not yet presented the recovery rates at all. I have found that the recovery values of cysteine/Cys and methionine/Met from the ODS column are higher by using acidic phosphate buffer (pH 2.0) than by using 0.1% TFA (please see file; PTH Roma RP-HPLC).

By the way, quantitative protein-expression study should be performed by the quantitative PDMD (Protein-Direct-Microsequencing-Deciphering) method (please see files; HepG2 fucoidan and JMBT alopecia).

Hi,

Interesting question!

I think it is impossible to extract quantitative information on potential differentially expressed proteins from the data you've obtained.

However, I think the data you have should provide you some means in distinguishing relative levels of protein amounts semi-quantitatively.

Following some filtering based on data quality parameters, such as FDR (false discovery rate), #Peptides, #PSMs (peptide spectral matches) and #Unique Peptides, you could start to look for black/white differences between the groups of replicates you're studying. This means looking for proteins that are uniquely identified in 1/3 or 2/3 conditions (for example at least in 3 out of 4 replicates per condition).

Next, you can look for proteins that are identified in all groups (most replicates), but that have differing Pep Scores, #Peptides, #PSMs and/or #Unique Peptides, as these parameters can be considered derived (indirect) measures of protein quantity. In that context, it would help if you have some proteins that you know are present in the same amounts in all three conditions, to allow for some kind of normalization and to provide some insight into the variation in your analytical methodology.

Taken together, the set will not provide accurate quantitative data with the current experimental setup (for that you would need some kind of (stable isotope) labeled internal standard), but it could provide some idea of potential differentially expressed proteins.

Hope this helps!

Dear Kou Hayakawa many thanks for your explanations and papers.

Dear Eef Dirksen, thank you very much for your useful explanation.

I was wondering if I can obtain protein concentrations from counting the number of peptide-spectrum matches for each protein from my dataset (based on Choi et al. 2015 paper that mentioned “Spectral counting has been a popular label-free quantification method due to the ease with which the data can be obtained by counting the number of peptide-spectrum matches for each protein from MS/MS search results. Spectral counting has been shown to correlate well with known protein concentrations”.)

https://www.ncbi.nlm.nih.gov/pubmed/26254008

Also, I would appreciate if you could advise me that, in general, in order for identification of differentially expressed proteins, absolute quantification is better or relative quantification?

Thanks a lot.

Hi,

No, it is not possible to deduce protein concentrations by counting the number of spectral matches; this parameter is merely (at best) indicative of the amount of a protein present: if there is 'more' of a protein in sample B compared to sample A, you expect to find 'more' spectral matches for that particular protein in sample B when compared to sample A.

Having said that, I think when considering differentially expressed proteins, relative quantitation (using either labeled or unlabeled approaches, depending on the accuracy you're after) should suffice.

Hi Eef,

Thank you so much for your help.