Talk:Proteomics
And while I'm at it: 1) its not just having a difference in protein expression between different cells, its having a difference in the protein compliment of the cell; different proteins are present in different cells, and at different amount, and even the activity of the proteins can also be different. 2: we are looking at how proteins interact with each other and with the genome to produce a phenotype that is appropriate for the environmental conditions that the cell or organism finds itself in. 3: I suggest that we not get into a discussion of what a gene is when discussing the proteome. I think that our traditional definition of a gene is pretty troublesome and will probably be revised over the next few years. We can say that there are fewer transcribed regions that code for proteins than was expected (although I think 22k might be low).
Then there is a point on good grammar; bulleted points should use the same kind of verb. for instance, this is what is in the article:
key technology
- 1-D electrophoresis and 2-D electrophoresis are for the separation and visualisation of proteins.
- To identify and characterise proteins mass spectrometry, X-ray crystallography, and NMR are used.
- To characterise protein-protein interactions, a number of chromatography techniques are used especially affinity chromatography. Protein expression systems like the yeast two-hybrid and fluorescence resonance energy transfer (FRET) can also be used to characterise protein-protein interactions.
this is what it should be
Key technologies
- 1 and 2 dimensional elect. are used to identify the relative mass of a protein and it isoelectic point.
- Xray crystalography and NMR are used to characterize the 3D structure of proteins.
- Tandem mass spect combined with reverse phase chromatography or 2D gel electrophoresis is used to identify and quantify the total protein found in cells
- Affinity chromatigraphy, yeast two hybrid techniques, and Surface Plasmon Resonance are used to identify protein-protein and protein-DNA binding reactions.
I think I'll just add that, but note the parallel verbs used in the 4 points.
srlasky 04:21, 2005 Jan 11 (UTC)
Was the selection below deleted? The 3D structure of purified proteins is a small part of proteomics, although it is a part of proteomics, and we are probably going to be able to find out a whole lot more about proteins by comparing the structure than by comparing their sequences. For example, using round numbers, when we sequenced the Halobacterium NRC1 genome, there were more than 800 potential coding sequences that bore no sequence similarity to other known proteins (all the genes named with a vng in them are unidentified and are called vng... for Victor Ng, the lead scientist on the sequencing project). By folding the predicted translation product of these unknown genes using ab initio technqiues and then comparing the 3D predicted 3D strututres, we were able to identify the function of more than 200 of the unknowns. So these structural comparisons are important, but are only a small part of what we are getting from proteomics.
What I started to write about was the inclusion of the "environment" as being important in the proteome. the article now states that the proteome is "constantly changing through its biochemical interactions with the genome and the environment. " I'm not sure what that means. How does the proteome biochemically interact with the genome?? and what part of the environment changes the constitution of the proteome?
First, I just don't know what biochemical interaction takes place with the genome. When proteins bind to the genome, they may be phosphorylated, but changes in phosphorylation are probably done before or after binding to DNA by autophosphorylation or by phosphorylation by some other protein kinase: for instance, the JUN protein is phosphylated by the JNK protein and them then the P-JUN protein can bind to an AP1 site. Dephosphorylation by protein phospotases is another important post-translational modification, but neither phosphorylation nor de-phosphorylation are catalyzed by the genome. The genome pretty much is inert biochemically. Its structure is changed by the alkylation, phophorylation, or other modification of DNA binding proteins, but the genome isn't doing it.
Second, what environment causes the proteome to change? Environmental factors cause biochemical changes in the cells. The cells can respond by changing the post-translational structure of preexisting proteins, or environmental factors can cause a signaling cascade to change the expression of genes leading to a changed proteome, but its not really the "environment" that is changing the proteome. It is the cell responding to different conditions that leads to changing the proteome.
I think that there is a better way to put the express how the environment causes the proteome to change, and I think it might be appropriate to retain the structural, maybe using the kind of example I used above.
Before I do any of that, however, I'd like to see a comment on my misunderstanding of the term environment. It just sounds too wholistic the way its in there now.
thanks srlasky 03:43, 2005 Jan 11 (UTC)
This was deleted from the article by 64.230.7.80:
- Two major approaches to proteomics exist: the study of in-vivo samples and the synthesis of recombinant proteins. In the second instance, genetic engineering techniques are used to clone the DNA template for the protein being synthesized and to splice these gene into host cells, typically bacteria, which are made to express the protein in large scale.
- The protein then has to be extracted from the host cells and purified. Subsequently, the pure protein is submitted for crystallization (and then x-ray) or NMR for structural determination. NMR is not effective for large proteins.
- Proteomics is a greater challenge than genomics because the 3-dimensional geometry of proteins is critical in their function. It is important and challenging to preserve this geometry through all the steps described above.
some of this could be relocated to structural biology, but some of it probably deserves to be in this article. 64.230.7.80, please don't simply delete material, but move it to the relevant Talk page, or a more relevant article. We try to preserve as much good well-written information on wikipedia (as the above paragraphs are), and if it's misplaced it should be relocated not simply deleted. --Lexor|Talk 08:54, 24 Apr 2004 (UTC)
new "branches of proteomics"
I just added a new section on "branches of proteomics". This is my first major Wikipedia edit, so I'd appreciate feedback. The "proteomics" entry is very short now and I'm not happy with the layout. The "key technologies" section is very limited now, and I think it should be revised. What do you all think? Janbrogger 21:04, 21 Mar 2005 (UTC)
- I like your edits, As long as everything in key techs is coverd, go ahead and delete the key technologies section--nixie 22:54, 22 Mar 2005 (UTC)
Recent change to caption of protein pattern analyzer
I'm the person who uploaded Image:Protein pattern analyzer.jpg from cancer.gov, and I've reverted a change made by 67.170.82.217 to this picture's caption. Personally, I'm unqualified to assess the accuracy or efficacy of the ECAN Genesis 2000, and I'm aware that captions on NIH pictures have been known to contain inaccuracies; however, to see a user with a limited edit history so radically alter a block of text makes it seem suspect to me. Compare the descriptions:
- Original caption: ECAN Genesis 2000 robot preparing Ciphergen SELDI-TOF protein chips for proteomic pattern analysis. Highly accurate cancer prognosis can be accomplished by identifying protein patterns of specific cancers using this device.
- Caption after edit by 67.170.82.217 (changes shown in bold): ECAN Genesis 2000 robot preparing Ciphergen SELDI-TOF protein chips for proteomic pattern analysis. Although many have tried, only highly in-accurate cancer prognosis can be accomplished by identifying protein patterns of specific cancers using this device, and its use in any clinically relevant application or research program is highly questionable.
If discussion on this page indicates some consensus that the accuracy of protein pattern analyzers is questionable, then I'm perfectly okay with it. Please understand, 67.170.82.217, that I don't mean to merely disregard your change, but instead want to vet it given that such a significant change has been made with no references or supporting argument.
-Quintote 11:37, 4 October 2006 (UTC)
- I haven't really done any research on this but to my understanding the prognosis of cancer with proteomics has a real mixed history. In addition I believe that in particular some SELDI based research has proven to be unreproducible. A statement has been added about it this particular instrument being "the laughing stock of what is now considered the dark ages of proteomics". I don't think that is too far from the truth although a very unencyclopedic way of saying it. To my understanding it is a prime example of an over-hyped and over-promised technology platform that missed its mark by a long shot. There was also a highly cited and heralded MADLI-TOF study of ovarian cancer diagnosis back in about 1999 by Petricoin et al. in The Lancet that used some fancy neural networks or something which ended up being seriously flawed. This set off a series of studies that were really poorly validated that were designed for the same ends. I have not stayed up with the level of sucess but there are certainly ways to approach the same level of predictive power as traditional tests using very focused mass spec based proteomics but then you might not call it proteomics. In any case the "dilute and shoot" (unfocused, little sample prep) approach has generally been a failure. I would say that both statements above are off the mark but then again it is a caption. How about:
- ECAN Genesis 2000 robot preparing Ciphergen SELDI-TOF protein chips for proteomic pattern analysis. Cancer prognosis by identifying protein patterns of specific cancers using devices such as this is an emerging field with a mixed track record of success.
- Again I have not doen any real research on this but I beleive this is a more accurate and encyclopedic statement than either of the others.--Nick Y. 17:27, 4 October 2006 (UTC)
- I haven't really done any research on this but to my understanding the prognosis of cancer with proteomics has a real mixed history. In addition I believe that in particular some SELDI based research has proven to be unreproducible. A statement has been added about it this particular instrument being "the laughing stock of what is now considered the dark ages of proteomics". I don't think that is too far from the truth although a very unencyclopedic way of saying it. To my understanding it is a prime example of an over-hyped and over-promised technology platform that missed its mark by a long shot. There was also a highly cited and heralded MADLI-TOF study of ovarian cancer diagnosis back in about 1999 by Petricoin et al. in The Lancet that used some fancy neural networks or something which ended up being seriously flawed. This set off a series of studies that were really poorly validated that were designed for the same ends. I have not stayed up with the level of sucess but there are certainly ways to approach the same level of predictive power as traditional tests using very focused mass spec based proteomics but then you might not call it proteomics. In any case the "dilute and shoot" (unfocused, little sample prep) approach has generally been a failure. I would say that both statements above are off the mark but then again it is a caption. How about:
- Thanks for giving this some insight, Nick. My sole concern is that the quality of Wikipedia is raised or maintained. Bold edits should be encouraged, but not at the expense of verifiability. I'm not hung up on getting references to peer-reviewed publications stating this, though obviously that's the desired goal; I just want a credible-sounding rationale for NPOV wording, and you provided that. -Quintote 03:28, 5 October 2006 (UTC)
I partly agree with the criticism of the SELDI-TOF MS platform, the technology has low sensitivity and low reproducibility. However, there is now a growing research area within proteomics, which is sometimes referred to as "MALDI-TOF MS protein profiling". This research uses not only SELDI-TOF MS but also the similar platform from Bruker-Daltonics (here samples are prepared by magnetic beads) as well as several other variants combining sample treatment and MALDI-TOF MS quantification. Other biotech companies are currently releasing similar platforms (fx Perkin Elmer). To state that this research has generally been a failure is clearly wrong.
Often this kind of research identifies highly abundant fragments of plasma proteins in serum for potential use in clinical diagnostics. The actual clinical impact is yet unknown.
I suggest therefore removing the reference to Ciphergen Biosystems, since it is only one among several MALDI-TOF MS protein profiling platforms. In stead include a description of "MALDI-TOF protein profiling".—The preceding unsigned comment was added by 80.164.177.134 (talk • contribs) 11:27, 23 October 2006.
- MALDI based systems are by far more reproducible than SELDI systems. There is nothing inherently wrong with either of these systems the issues of the past had more to do with the claims made than anything else. MALDI is an absolutle wonderful technology but is not the most reporducible thing in the world and requires expert involvement to get reproducibility out of it. There is a major difference between instrument reproducibility and assay reproducibility. Maybe you need to run the assay three times under different different conditions averaging many laser shots and you can get very robust indicators. Another problem people had in the past was the choice of sample preparation schemes that tended to leave way too much abundant proteins around masking the signal from the more relevant proteins. There is no doubt in my mind that protein profiling by mass spectrometry is a viable approach to diagnostics and there are examples of successes to prove this but there are also examples of failures. The technology is fundamentally sound but requires well designed thoughtful and even sceptical approaches. I would agree that the ciphergen SELDI system should be removed as the poster boy for this type of approach. Be bold. The change is welcome.--Nick Y. 17:38, 23 October 2006 (UTC)
- I think that sounds great as well. The picture, however, is of a machine preparing Ciphergen SELDI-TOF protein chips. If this looks exactly like a MALDI-based system, then I'd be okay with changing the caption, but I'd be concerned about misrepresenting the picture. However good or bad SELDI-TOF profiling may be, that's what's in the picture. -Quintote 23:43, 23 October 2006 (UTC)
We agree that the picture of SELDI should be removed. It would be better to show a picture of another, less controversial, MALDI-TOF MS platform. In an article about cars you would show a commen and well tested car, rigth? Find a picture of a Bruker Daltonics MALDI instrument.
However, the statement "MALDI based systems are by far more reproducible than SELDI systems." is wrong. The average CV of the m/z is below 0.1% with SELDI as it is with all current MALDI protein profiling platform. Yes, you can get MALDI platform that are 10 times more mass accurate but these cannot be used in protein profiling, where a robust instrument design is desireable. And more importantly, the m/z ratio is of less importance in protein profiling. Here it is the reproducibility of the "peak intensity" that is of interest. No other MALDI platform has been shown to be more reproducible with respect to the peak intensity than SELDI (however, the variation is in general very high for all platforms). The problem in MALDI is not the instrument, the general problem is the variation inherent in the matrix-protein co-crystallization step, which is common to all MALDI platforms. The often heard statement that the SELDI instrument is a poor MALDI instrument is clearly wrong, and is argued by people who are not working with protein profiling but with other types of MS research.
- The m/z "reproducibility" (accuracy) is dependent on the mass analyzer not the ionization technology, although in some systems the interface between the two is important (delayed extraction, etc.). Most common systems are TOF based. SOme of these newer TOF are quite impressive. You throw an FTICR on the back end and you have thousands or times more m/z resolution. Yes, MALDI signal intensity is largely dependent on the MALDI plate preparation (co-crystalization). I was including that in "instrumental parameters" as opposed to assay reproducibility. An assay can be developed around this reproducibility issue giving an accurate diagnostic. This, however, is a major issue and regulators are generally hesitant about diagnostics on such platforms. I said before there is nothing inherently wrong with SELDI. My statement about MALDI being more reproducible was more directed at a historical perspective. I should have said "have been". To my understanding the failures of the SELDI system had more to do with experiment design and the claims being made. Again I have not thoroughly researched this, but that is my understanding. Please give greater insight to this if you have some. I have seen excellent reproducibility out of MALDI systems but in the hands of a single expert, under well controlled and designed conditions. There is still an art to it. Internal validation with check samples etc. is important to any robust method. You must agree that SELDI has developed a negative reputation deservingly or not based on some high profile failures. MALDI has a good reputation which is often even over estimated, despite some major failures. Both techniques are often misunderstood. It is easy to reach false conclusions and design worthless experiments on both if the nature of the techniques are not understood. I really don't know that much about SELDI and would love for some one to write an article about it. Surface-enhanced laser desorption/ionization--Nick Y. 17:13, 24 October 2006 (UTC)
I think we agree after all. I am perhaps a bit more sceptical about reducing the reproducibility of the peak intensity simply by using well controlled and well designed experimental set up. I have seen studies of fully automated MALDI-TOF MS protein profiling experiments (fx sample loading and matrix application by robots), which in my mind do not show significantly improved reproducibility as compared to manual studies. However, this is difficult to conclude about, since the variation differs dramatically between individual protein peaks (presumably due to the physical/chemical properties of the primary sequence of individual proteins?). I think we need research in how to reduce the empirical nature of the crystallization step. There should a place in the history books for the person who solves that problem?
Great idea about a SELDI article.