What is epigenetics?

It's a question that I am often asked. The answer is complicated. Since, in my opinion, the viewpoint of the field shifts based on where the next sexy science that can generate money is coming from. 


So lets start with the word epigenetics and its semantic meaning. If we start by breaking down the word, we can see what the word and field has come to mean in the last twenty years. The first part epi is from Greek meaning above. Genetics....well realistically it is a galaxy of smaller fields dedicated to studying what genes are, how they are involved in disease and development, and how genes regulate each other. 

So epigenetics is the study of processes above genetics.........which is where the etiology fails to be useful. Hence the confusion amongst scientists and the public at large.


In a practical sense epigenetics is a in-depth look at how genes regulate each other. Genes are, at their strictest definition, the precursors to the proteins that perform (most of) the jobs that make life possible. 

While genes are the stars of the show, in terms of regulation of biological processes, they are the least interesting part of the genome. The interesting part is the so-called junk DNA or more accurately non-coding DNA. If genes are the stars, then non-coding DNA are the role players and the scenery that move the show forward. 


The dance of how these proteins and modifications is choregraphed is epigenetics. 


Non-coding DNA regions control which genes are expressed and when they are expressed. This occurs through RNA molecules, binding of proteins and enzymatic modification of these proteins. Its more complex that what I am highlighting but the core message here is that Epigenetics is about controlling access to the genome by RNA and proteins through the non-coding regions of genomic DNA. 


Epigenetic processes are the key to providing organisms the flexibility to respond to the environment. So what about these processes make them so important?  As with any regulatory process-it depends. 


The best analogy is the "genome as the book of life". If genes are the words by which an organism is "made" then epigenetics is everything else in the book. At the lowest level it is the sentence structure that allows the words to have meaning. At the highest level it is the chapter order that gives the story a linear order. Unlike a real book each tissue in the body can shuffle each of these elements on the fly....a choose your own adventure book based on cell type. Further flexibility arises from how each individual cell interprets the story that it is given.

While the book analogy is interesting and useful, the beauty of the system in my opinion is that all of this is managed through biochemistry: small differences in enzyme kinetics and subtle changes in protein binding. This biochemistry leads to a wide variety of markers that can be used to denote parts of the genome.


The cell uses different types of markers for each of the particular categories; grammar, pages, chapters. Each of the different markers that the cell can use for these categories has varying ease of use. The ease of use is a function of the biochemical processes by which the cell adds or remove the markers. If this all seems like overkill just to express a gene...it sort of is. Adding layers and differing ease of use allows the cell to add very tight regulatory control. This allows the cell to mix and match different markers to make bookmarks or highlight favorite passages, often used paragraph, etc.  

In general the chapter markers are direct methylation of the DNA. As the name suggests it is the addition (or subtraction) of methyl groups to DNA. This alters the affinity of DNA to a subset of proteins that occlude the DNA-basically hiding the genes. Removing the methyl groups abolishes this occlusion.

Sentence structure this is more complicated but in general there are 2 types of post translation modifications that are most commonly seen as having a definitive role in this process. This usually involves the structural proteins that surrond DNA. These proteins are the histones and they allow 2 linear metres of DNA to be packed a volume of 0.00001 metres. An amazing feat in and of itself!

Histones can be modified in a wide variety of ways too numerous to mention here. The combinatorial placement of these proteins and modifications on DNA allows for a very precise grammar to be used. So precise that cells can communicate exactly which protein should be expressed to their neighbor. Given that there may be as many as 31 thousand that is quite impressive. What are histones? that is another story for another post. The bottom line here is that these proteins control access to the DNA. The modifications on these proteins act to either loosen the structure and increase access or glue the proteins together blocking access.

So what you say? Well the answer is variation: variety in cells (brain VS muscle), variations between twins, variation in cell response, variety in drug response, variation in disease. Epigenetics is the root cause of variation at every level: species, organism, tissue, cells.

So until we understand how this biochemical signature is modified we can't really understand how cancers vary from person to person or in a bigger picture how we retain biodiversity. 

Health IT and clinical research

I recently returned from the E-Health Canada conference in Vancouver. I was there as an analyst for Info-Tech research group. I spoke about secure use of consumer devices in healthcare and the potential of cloud computing as a flexible model to deal with CoIT.


I was pleasantly surprised by the IT knowledge level of the nurses and doctors that attended the conference. Why was I surprised in this age of consumer devices and self service tech? Well I spend a fair amount of my time talking to and about IT departments. The impression that I get from many of them is that while most people can set up their email on their phone they remain largely clueless about the actual tech itself.  


Well, there is certainly a core of nurses and doctors that understand the tech and have really great ideas for how to make it work better in the context of healthcare delivery. I was left with the feeling that many remain frustrated with the current solutions and the pace that E-health is moving forward. The major areas of frustration were around content delivery and system upgrade for usability. I would summarize it as “You must do something; protect the data but be flexible on the device used”. Technology should allow doctors to spend more time looking and talking to patients not with their nose buried in a tablet. Just placing a tablet in the doctors’ hands doesn’t mean it will lead to increased use of the IT solutions for healthcare.


While some may point to this as a technology problem, it was clear from talking to the vendors that the solutions available today can meet the demands that hospitals are placing on IT. As someone with a past on the research and has dabbled in clinical research it was refreshing to know that there is a variety of solutions out there to make clinical research easier. What was interesting was how similar healthIT problems are to many other industries. In my opinion the issue is now about getting the best out of the tech not when will the tech exist.


 In other words its about getting the users and the admins on the same page. 


As someone with a deep passion about Rare diseases and use of high level biomedical research its somewhat frustrating that the system that is in use today is so antiquated. The upgrades available today could add so much intelligence to how we treat Rare diseases in particular. 


The new areas of stem cell therapy and epigenetics hold a HUGE promise if we can understand the relationship between disease and biology in these patients. Since they are so rare at any given location it is imperative that we have a way to share the data that is safe enough to share patient history across borders.

Open access and generating senior scientist buy-in

There is a vibrant, intelligent but completely impractical debate happening right now around publishing scientific papers just use #openaccess to see the volume of twitter posts. The concepts are great-faster dissemination of information, cleaner peer review process, greater collaboration. 

My issue is the lack of reality check or way to bring it into real world use. Science at its heart is a glacier- cold, unworried, progressing forward in an unstoppable manner. It also has the inertia of literally hundreds of years and in general a highly conservative set of guidelines. What needs to be fought against is this idea that peer review requires a third party outside of the scientific community. The internet and the transparency that it brings makes a third party paid watch dog unnecessary, there are plenty of folks on the internet just looking to "yell gotcha".

There is some value to the conservative mindset that works well for society and Progress-the burden of proof. The conservative guidelines protect Science from making too many mistakes and jumping to conclusions based on the unexplained. It's what prevents Einstein's theory from being torn down by a faulty wire. It also means that the younger generation of scientist must bring ideas to the hallowed halls of old science and prove that it will work by bringing real world suggestions that will fix the interwoven problems of publication, attribution, grants, jobs, and tenure. 

Otherwise its just pissing in the wind and complaining. Not my idea of what scientists do. They come up with theories, models and explanations-that are then roundly torn  down by their peers and rebuilt into a better product.

With that said lets take a look at how open access and "non-publishing" publishing could work in the real world*. 

First some challenges that I see as the main drag on process:

1. Comparative analytics. There needs to be transparent metrics to gauge the value of the research to the larger community. This can be an issue since every scientist does the most important research to the world.
2. Control over content-this may seem trivial but if I am the principal investigator I'm not sure I want every post-doc and grad student uploading their crappy, blurry images. The metadata surrounding who gets tagged, whats get tagged and the terminology is vital to ensuring that the data can be reviewed by everyone who may be interested. 
3. Clear lines of permissions-What role do collaborators play in making this public who gets to post it? where is it hosted? The university still has some co-ownership. This is not something that can be decided afterwards it has implications for grants, etc.
4. What about intellectual property? I'm all for open collaboration and sharing data with academics but what guarantees are there that pharma and biotech will play by these rules? The beauty of public research from the grantor's(read government) point of view is it maximises their investment. That is lost if [insert huge pharm company here] comes along and builds a drug based on that data and charge the public obscene amounts of money.
5. The content that is made public has to have a finished look. This can't be just thought vomit. It must have some clarity of thought and citations. Put some thought into it! Distill by putting into context, why should I care? how does it support or counter the current models. Proper citation through hyperlinks. It should be peer review not crowd sourced editing.
6. Control over comments. Comments can't be removed just because someone says your data quality sucks. This has to be transparent warts and all. You have to take reviewers suggestions of more experiments seriously. As someone who has reviewed for a variety of journals (top five to lower tier) nothing is more frustrating than taking the time to review and having someone completely disregard it. 
7. Buy-in from senior scientists. Science is largely a oral history, the reality is that for most methods just having the paper is useless for understanding how to do it or where the problems can arise. The insights from senior scientists that have seen it all are required for this to be truly revolutionary. 
8. Buy-in from Administrators. I for one do not believe that open publishing will be any less time consuming or cheap for the scientist. Someone will have to maintain the database for the content and ensure that there is enough capacity for the videos, excel files and what not that will be made available. This will have to be maintained for decades with back-ups, etc. Someone will have to know where and what there content there is AND when is should be taken down and replaced. Right now the university covers those costs for each departments website, unless there is a clear benefit i.e. grant money, donors, clarity on which faculty members are doing well and in a perfect world which require more help.

Now the potential solution:

A dedicated website where each lab publishes its own data. Personally I don't be believe that all data should be available but some labs and branches of science believe that is best. Allow the system to have some flexibility, some fields are inherently more competitive and technically nuanced than others. Scientists need to retain the ability to check data quality, accuracy and potentially fundamental flaws in experimental setup as a lab group prior to making it public. 

I like figshare and creative commons and all of the really great tools that are coming at breakneck speed. I love the idea of posting of all of the data but I truly believe that my job as a scientist is to analyse the data not just find/acquire it to send to others. If open access publication does not include this it will set back cancer and other highly nuanced fields in biomedical sciences years. These fields have moved at breakneck speed because such a premium is placed succinct analysis by the publishers. While I do not believe publishers should be the gatekeepers, I do not want to lose the analysis of data for the sake of speed of publication. 

The solution: most labs have university owned/manged websites where the Principal Investigator (aka PI, professor i.e. the person who's tuckus is on the line) owns the admin rights. These should become more of a real world home for the publication and sharing of lab data. It exists and some labs do a great job of updating it with content as it gets published. Everyone else needs to get on board bring it into the modern age with appropriate tagging and labels to ensure that it can be found through search engines.

PIs need to retain control, they are the ones that will be held accountable if the "published" result that shows a new cure for cancer that turns out a contaminant. The pervasive nature of the Internet means that the media has access to the data and need to hype themselves as getting the most interesting story. "Never let the truth get in the way of a good story" is a truism for more and more journalists. Accidents happen and while I'm alright with being embarrassed by my peers figuring it out, I'm not alright with it spinning into a worldwide story and having to explain that it was an "oops" to the public. 

Main point: Use the established website, each PI has admin rights to remove. Give senior lab members the ability to publish concise analysis with appropriate figures and links to the whole data set with metadata and clear descriptions. As part of the mentorship for new post-docs and graduates students training on what is considered an acceptable level of proof prior to making the data public. Laboratories are still training grounds, as someone who has trained students and post-docs the peer review process allows young scientists looking to move up to the majors an idea of what good science is-this cannot be lost by opening up publication. I love the citizen science movement but at the end of the day, like anything, there has always been a difference between someone who does something as a hobby and someone who has the discipline, passion and willpower to dedicate their life to a subject. Training and the culture of science has to be part of it-science is about justifying your opinion and the quality of data. 

Peer review isn't broken, the publishing models are, let's not throw the baby out with the bath water. The university controlled site allows for clear rules of engagement for pharma, media and allows for the level of control that PIs, chairs need to ensure that crappy data doesn't not spiral out of control into a scandal. The departmental chair and grant study groups can look at the metrics; website views, re-links, etc to allow flexibility into the systems for review whether it be tenure, grants or something else that no one has thought of. Open access will be a failure if it does not give everyone involved with the industry (yes its an industry! get over it people). It's not perfect but it can be piloted in a way that senior scientists from the core Cell, Nature and Science author pool can at least talk about. I think that many of the ideas that are being bandied about are much better than this for science as a whole and ultimately will be the long term solution. 

That being said I have yet to see an idea that any of the top level Cancer, Stem Cell, etc scientists will buy into. This is not a group to dismiss, they may not be the majority but they represent the main attraction for why scientists will not give up on the Elsevier or any other for profit publisher. They also are the presidents or senior leadership of some of the most influential universities in the world (Caltech, Rockefeller U., Memorial Sloan-Kettering, Max-Planck, etc). As a final reason to get them on-board they also are on the grant study groups and a variety of other activities that effect all levels of at least biomedical sciences.

At the end of the day the risk posed by publishing incorrect data needs to be balanced with greater access and conversation about what can be done next. Please comment as you see appropriate.

*Disclaimer-I only have experience with a limited number of institutes(eight) so I do not know if any of this is applicable widely. I have no idea if the issues that are bringing up are universal or limited to the institutes that I have worked at.

Rare diseases and open access.

Ive found myself completely mesmerized by the open access/open science debate. As a recovering bench scientist, it has made me think about a variety of things but one that is really interesting is the implications for Rare disease research and speed of turning great benchwork into viable drug targets. Ill deal with the larger debate on open access separately but I wanted to put forward something today(Feb 28th 2015 Rare disease day). 

2016 update: In my opinion not much has change in Rare diseases in the last year. There have been some moves forward but like anything in the drug and/or therapeutic research- it is time consuming. I hope that the silence is because we are getting to the point where folks have rolled up there sleeves and are working not talking. 

I am really excited about the prospects for increasing the speed that potential drug targets can go from bench to bedside. The new technologies (gene sequencing, clinical data) can provide faster turn around time through efficient data sharing and new genomics technology. The real potential pay off is through new clinical data that will be available once EMR is implemented widely. The value of that much data combined with the new genome sequencing technologies can really provide some much needed guidance about the genotype phenotype relationships that may link certain rare diseases. I say may since it will really come down to data quality and wide dissemination of that data. Getting clinical data into the hands of molecular biologists and biochemist who can do the bench research is vital to drug design. 

2014 Update: With the roll-out of ACA starting to happen and the FDA crackdown on 23andMe. The landscape for studying and curing Rare Diseases just got a little better. For more information on the 23andMe nonsense there is plenty of information on the imbroglio but this one from the Huffington Post is the least sensationalist. My opinion is that the FDA made a decision based on the specific businesses lack of response it is not an indictment of consumer genetics or any paternalistic over-reach. Mathew Herper has a really great analysis of the stupidity and or hubris that 23andMe showed.  The Global Genes Project has a nice blog on the relationship between Rare diseases and ACA. 

The bad news is that the sequester has set research back years if not decades and may have very well rob a whole generation of scientists of their careers (this author included). Tom Ulrich of Boston Children Hospital has a nice blog on the subject.

\
2015 Update: The new interesting initiative is precision medicine in US. I really am proud of the way Global Genes Project is coming along. I was briefly involved when Nicole Boice started the initiative. I look forward seeing how it continues to grow in 2015. I think it does a great job of keeping the conversation on awareness and providing a site to aggregate "best practices" for the Rare disease community. 

I really hope that the continued access to healthcare that we started to see in 2014 continues. The key will be what do we do with the basic information that clinics gather about their rare diseases patients? How do we make that shareable across clinics. This in my opinion will be the key to consistent diagnosis and clear symptons, which will then better inform scientists of which genes contribute to the phenotype. This is the basis of drug discovery and treatments. 

Right now is a real nexus of information due to the convergence of new technologies with "new" fields of studies. Epigenetics is the study of how and why genes get turned, the best analogy is: if the whole genome is the book of life, genes are the words and epigenetics is the sentence, paragraph and chapter structure that gives the words meaning. The other area is off-shoot of stem cell research; induced pluripotent stem cells (iPSCs). iPSCs as the name implies are induced to become stem cells from a variety of other cell types, the most clinically relevant being skin and blood. While the debate rages iPSCs and their value for replacing non-working cells with new ones [regenerative medicine] one thing that is not in doubt is the power of these cells for modeling disease. iPSCs can be made from patient samples and then shared with other researchers. This may seem trivial but the more people looking at the same model the quicker the core problem can be found. If done right the sharing of the iPSCs to researchers who use different techniques (biochemists, molecular biologists, cancer, etc) will provide a 360 degree view of the disease. 

Update 2014: Unfortunately it seems that iPSC research is becoming marred with scandal. The new "most promising" discovering may be "less real" than one would hope.....Paul Knoepler has a blog on the subject. BTW if you have any interest in stem cells you should follow Knoepler's blog he is an excellent writer and a top notch scientist.

Update 2015: I think we are past the really bad period, unfortunately it has also diminished the enthusiasm for iPSCS as models. Although I am not surprised, it has recently been shown that iPSCs form different sub-types of cells based on their tissue of origin (see here for neural and here for heart). This seriously limits the usefulness of iPSCs for drug discovery and would just exacerbate the reproducibility issues that are plaguing science in general but particularly stem cell research. 

Once this happens it's likely that links will be found that can make drug discovery and testing palatable for biotech and big pharma. Drug discovery is expensive but if the community can gather enough information about the molecular and biochemical characteristics of rare diseases then the existing "orphan drugs" can be tested against the characteristics rather than any single disease. 

Update 2015: The orphan drug area is one where we are starting to see movement. The recent announcement by the CF foundation recieving $3.3B for the patent rights to Kalydeco. It is an interesting approach that should be considered by any rare diseases group looking to expand support and the potential therapies for their disease. 

As always the caution is who should get the money, how do you ensure that the cost of the drug to sufferers is appropriate? If the foundation funds the study (in part) do they have an obligation to ensure that the cost of teh therapy be reasonable to the average person?

The elephant in the room is of course paying for all of this. Scientists need to be able to publish to get grants to pay for post docs and reagents. While there is some money available from disease foundations but it doesn't cover all the costs that a lab needs to run. That is the job of the NIH. However their mandate really requires that grants are given out based on WIDE applicability of the research and the grantee's history of research in that area. Unfortunately this model does not serve the rare community very well nor does it foster the wide range of scientific endeavor. There hundreds of examples where a rare disease has lead to unique insight into a biological pathway that was key to some cancer or other disease. 

Update 2014: Rare disease research will survive but we need to start to fast track new funding models that focus on highly innovative projects. We know what hasn't worked we need some research that is different.

Update 2015: Unfortunately I can't say there has been too much movement on this. Frankly scientific funding is horrible right now. I think for rare disease foundations there is an opportunity to foster young scientists to be advocates and invested in their disease but this requires a new way of thinking about how to fund rather then WHAT to fund.