Add Proteomics to ENCODE?

[Doug]

I just read interesting articles in Nature Biotechnology and ProteoMonitor discussing the possibility of adding a proteomics component to the ENCODE project (Encyclopedia of DNA Elements). They discuss the advantages (e.g. knowing if variants affect protein levels or function) and the disadvantages (e.g. proteomics may not be a mature enough technology for this integration).

Uniting ENCODE with genome-wide proteomics

Human Proteome Project Organizers Propose Role for Group in ENCODE Initiative

Tell us what you think.

Note - Voting in the poll is anonymous

[gadsouza]

I think this is quite a daunting task if one imagine all the different plethora of data formats and routines for spectra processing, interpretation and identification. I think it has already been discussed elsewhere about the issues with metadata repositories in proteomics.

I would only support or even trust such efforts for proteomics if certain rules apply, such as only data produced with instruments X, data processed with software Y with parameters Z etc are eligible to submission.

[Santosh]

Peptide data could be represented irrespective of the platform. One could mention how many experiments/platforms a peptide was identified. However, certain rules holds true for proteomics data incorporation such as FDR cut-off etc.

[Christopher]

There was an interesting paper that came out in Journal of Proteome research a couple of weeks ago on the "Current Status of the Human Peptide Atlas Project". I think it provides some interesting insight into some of the challenges there are for amalgamating proteomics data from a wide range of projects and platforms while controlling quality via FDR.

There were also quite a few papers this month in the same journal now talking about the chromosome-centric human proteome project...and this to me kind of has an ENCODE type aura around it. I don't know if I am a believer in these types of projects yet...especially for proteomics when we are just not at the level that genome and transcriptome studies are with sequencing technology.

[aky]

I agree with Christopher. There are too many issues to tackle before proteomics data integration from different labs, instruments, softwares, analysis pipelines, statistical tests, become practically feasible. The community has to come up with better and robust tools for doing that before we can think of such meta level analyses.

This doesn't mean it is not possible. It just would be too much work and maybe comparatively lesser output (salience, interest of other communities etc, impact on a general researcher). ENCODE is already criticized for telling us what we already knew. For whatever is novel, there is a debate on the semantics of what is called "functional". It seems practically difficult to get that kind of funds as well in proteomics. Maybe within a few years (I don't know how many), we can reach the level of sophistication that can allow multiple labs concurrently working on different aspects of a proteome-wide ENCODE-like project with community participation including MS-vendors, biologists, mass spectrometrists, bioinformaticians and statisticians. Just hoping, that time comes sooner for proteomics to blossom.

[gadsouza]

Peptide data could be represented irrespective of the platform. One could mention how many experiments/platforms a peptide was identified.

While this would work fine for abundant proteins/peptides, I would still be skeptical about reports of rare isoforms or SNPs. Unless they are reported under very stringent criteria (more stringent than what I normally see in the average proteomics article)