0
$\begingroup$

I want to use either of Proteomics or Transcriptomics data for integrating it into my kinetic model. Before proceeding, I want to know what are the advantages of using either of them so that I could make an informed decision on it!

Many studies have shown that the best we can do is integrating both transcriptomics and proteomics data with our kinetic model, but I've some time constraints and have to proceed with only one of those.

My effort and findings: I've found from discussions with researchers that gathering transcriptomics data has an amplification step which increases the chance of finding a particular one whereas gathering proteomics data has no such step but has fragmentation and then rejoining which creates many problems(such as splice variants etc) and thus leads to a loss of data. But a PostDoc told me that even after the loss of that data, I'll get more information from Proteomics data.

I want to know such type of points and want to know if these are valid or not!

$\endgroup$
3
  • $\begingroup$ You want to compare two different techniques but do you have data from both of them for your model? Theoretically they can be whatever, but then in your experiment you can get unusable data. Also it seems like you have two implementations of the model, so it might be influence the accuracy of the model. Did you try the model with simulated data of both types? (or with freely available information on the web) $\endgroup$ – llrs Jul 15 '19 at 8:41
  • $\begingroup$ I am yet to gather the data(either transcriptomics or proteomics) for my model. I have the list of genes ready for which I want it. I have only one implementation of the model only. I haven't tried the simulation with any data as I don't have it yet. I want to make an informed decision on it. Please suggest @llrs $\endgroup$ – Akhil Verma Jul 15 '19 at 9:27
  • $\begingroup$ I did some research and have put down my findings. Please add any suggestions that you want. @llrs $\endgroup$ – Akhil Verma Jul 15 '19 at 10:11
0
$\begingroup$

Ultimately, proteomics data can actually measure the actual concentrations of the enzymes that you are interested, while transcriptomics only measures their rates of change due to ongoing transcription from mRNA.

Rates of change are often treated as proxy for concentrations. This makes sense when you have a constant degradation/dilution environment and are near steady state, since then the rate of change is proportional to the converged concentration. If you're interested in more dynamic environments or believe that modulation of degradation or sequestration might play a significant role in a system, however, that assumption breaks done.

Thus, if you have to choose between proteomics and transcriptomics data, I would suggest proteomics.

As a final note, however, since the link suggests you're interested in metabolites, however, have you consider going directly to metabolomics data? That would let you really get direct information about the particular species of interests.

$\endgroup$
0
$\begingroup$

This is what I found from doing some research. Comments are welcome at any point!

  • Capture percentage for data gathering:

    • Transcriptomics Data: There's an amplification step in Transcriptomics data gathering methods. Hence, it's possible to capture almost the totality of the Transcriptome using those methods(scRNA seq methods, Nanopore tech, Spatial transcriptomics, etc)

    • Proteomics Data: There's no amplification step in Proteomics data gathering methods. Those methods have fragmentation and a rejoining step which leads to a loss of the data.

  • Scalability of methods:

    • Transcriptomics Data: Scalable methods available

    • Proteomics Data: Less scalable methods for protein studies

  • Reference availability:

    • Transcriptomics Data: Fully annotated references available on consortiums such as ensembl biomart, etc

    • Proteomics Data: Universal and Comprehensive Human Proteome reference is still in question and because of the incomplete and inaccurate references, much of the newly generated data is being rendered useless

  • Uniformity

    • Transcriptomics Data: Lots of publications on pipelines, and their benchmarking available

    • Proteomics Data: Lack of uniformity across labs/research groups and lack of related literature lead to differences in protein fragmentation and solubilization and differences in algorithms to run analyses

  • Technical Bias

    • Transcriptomics Data: Many methods available to tackle cell-bias, noise in data, and batch effects, so as to capture maximum biological variability

    • Proteomics Data: Cell-wide Mass spectrometry has a bias towards the identification of peptides with higher concentration or contamination from other experiments

$\endgroup$
5
  • 2
    $\begingroup$ On the plus side, proteomics is actually measuring what's most relevant to your model. $\endgroup$ – Devon Ryan Jul 15 '19 at 10:12
  • 1
    $\begingroup$ @DevonRyan yes, sometimes I think that the field is focused too much on transcriptomics... I hope to see more usage (and learn more) on proteomics. On which side do you lean Akhil? BTW Did you take into account the experimental and bioinformatic experience as well as the cost of each to obtain (and store/process) the data you need? (just throwing ideas that are worth considering from a practical point of view) $\endgroup$ – llrs Jul 15 '19 at 10:23
  • $\begingroup$ I'm more inclined to use the proteomics data only but my PI has told me to go with whatever data will be of higher quality as then we can draw some inferences more confidently. Also, I didn't get your last point. Can you please elaborate on that? @llrs $\endgroup$ – Akhil Verma Jul 15 '19 at 13:44
  • $\begingroup$ Is anyone on your work (lab?) that has done RNA extractions for RNAseq or anyone that have done the preparation for mass spectrometry? Assuming you have them. How much would it cost to have the same quality of data from both techniques? Once you have the data, is there anyone that has the knowledge and resources to process it? In RNAseq you need to align to the reference (computationally expensive and time consuming), normalize/prepare it for your model while store all the data and the original files you got the fastq.gz. How do you plan to do that? $\endgroup$ – llrs Jul 15 '19 at 14:07
  • $\begingroup$ My lab is purely a computational one. We don't have the resources or people to perform the said experiments. So, my PI has asked me to gather the data from the existing literature. In the same way, we gathered the metabolites and flux data from the existing literature on Mass spectrometry and have used it to construct a kinetic model. Just to tell you I'm very well familiarised with the reference alignment, normalization, etc as I've done an internship in the scRNA seq analysis, but I don't think it'll be useful to me here in anyway. @llrs $\endgroup$ – Akhil Verma Jul 15 '19 at 21:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.