The Heroes of CRISPR
cell.com
cell.com
https://pubpeer.com/publications/D400145518C0A557E9A79F7BB20...
1. https://twitter.com/stephenfloor/status/688466159021064192
Lander's article is a completely different beast. I highly doubt Cell cold called Eric Lander and asked him to write a campfire history of CRISPR.
It is clear the Cell has a liberal COI policy in any case. I can't find COI mentioned anywhere, even to say that there aren't any.
Finally, the narrative underscores that scientific
breakthroughs are rarely eureka moments. They are typically
ensemble acts, played out over a decade or more, in which
the cast becomes part of something greater than what any
one of them could do alone.
In computer science terms, the method scientists use to discover and characterize genes (CRISPR, etc) seems akin to studying the assembly of a program, occasionally splicing chunks of assembly into other programs, and observing changes in the output after running. This sounds like a huge slog, and it is amazing that the method works.
One interesting thing from the article was that, even after determining the function of CRISPR, scientists still had difficulty in understanding the mechanical/chemical means of its behavior: the gene is a black box, and its expression can only be deduced when observing its effects after running.
Have there been attempts to characterize genes on a more basic level, eg by modelling how a given nucleotide sequence encodes a protein, and deducing its function from there? Is our understanding of protein folding still inadequate? Or would reconstructing a protein's physical structure still not give us enough insight into its intended function?
Modeling is prevalent and helps. Deducing function with any accuracy requires much better and more comprehensive models of cellular biochemistry, however. Those models lie decades of work from here, meaning that altering genes and watching the outcome in mammalian cells and individuals will be the primary mode of exploration and validation for a while yet.
Thanks to programming, a lot of this is automated.
In Biology, you can use programmable robotics arm that perform thousands of experiments (high-throughput screening). So for CRISPR, suppose you want to test the thousand genetic variation identified previously computationally for a particular cancer; you can take a cancer cell-line, apply gene-editing targeted at each of those sites each in a slot in a 96 plate-well and observe if the cancer cells' tumor growth. These 96 plate-wells are then fed into image analysis program to quantify the tumor growth.
>Is our understanding of protein folding still inadequate? Or would reconstructing a protein's physical structure still not give us enough insight into its intended function?
The field that simulate protein-to-protein simulation is called MD (molecular dynamics) simulation. The issue is the simulation is really complex. So you start at the molecular level, whether one protein can attach to another protein's end like Lego's, then you have to account for the individual cellular level, called cell circuits (modeling an individual cell like a logic circuit, different genes producing proteins that regulate one another), then you have to account for the multicellular interactions (how cells influence another).
Instead of trying to simulate everything, you have people comparing and studying things at different level.
Folding the protein is only one part of it. This part, while not "solved", has seen an enormous amount of progress in the past few years, such that we can often predict the coarse-grained 3D structure of a protein, although not the fine details in most cases.
Understanding the enzymatic action of proteins typically requires simulating chemistry at the quantum (bonds breaking and forming) level around the active site.
Since biologists are more interested in getting results quickly, rather than solving the fundamentally hard problems (the former gets grant money more easily), and because structures that are similar tend to have similar functions, they tend to use structural homology- similarity to a protein of known function- to infer the function of a protein.
Although much of my prior work was designed to address the forward prediction problem, I have acknowledged that shortcuts produce more valuable data. And often times, that does involve finding a gene whose protein product behavior is cryptic, and then using structural homology, and other indirect methods, to refine the function of the protein.
As for the "computer science terms", I can mention that after working with large distributed systems for a long time, I treat debugging them a lot more like how biologists deal with cryptic proteins than trying to understand them from first principles. I often run "experiments" by injecting things into the distributed systemns, and monitoring them, much like scientists monitor proteins using fluorescence.
This is going to be next year's medical tourism, analogous to the progression of stem cell therapies from the turn of the century.
There are a few really obvious candidates, such as myostatin and follistatin to greatly increase muscle growth - and at least one person who has had that done already, c.f. BioViva. To do this all you need are the connections. Buy into a biotech startup company, make the arrangement with a lab in Mexico or Thailand and off you go. If you knew the right people, you could be on a plane tomorrow and the beneficiary of elective gene therapy the day afterwards.
It can't be overstated how easy CRISPR makes this. It is easier even than induced pluripotency, and that spread like wildfire through the labs when it emerged.
The more interesting thing to me is that there are probably a hundred less obvious, poorly studied, poorly followed up, but very interesting genetic alterations that could be done, and probably will be done largely outside the institutions of medical research. This is what happens when cost falls. Any single gene is fair game now. Want great resistance to ischemia/reperfusion injuries? Knock out PHD1 [1]. Want your aged liver to function as well as it did when you were young? Add more lysosomal receptors. [2] Want to have permanent operation of the benefits of fasting and exercise in the form of upregulated autophagy? Increase AMPK levels [3]. And so on and so forth through scores of genes. Maybe many of them will work as the studies suggest, maybe not.
[1]: http://www.vib.be/en/news/Pages/-VIB-researchers-discover-po...
[2]: http://www.nytimes.com/2009/10/06/science/06cell.html
[3]: http://www.salk.edu/news-release/how-the-cells-power-station...
As a layman, this would seem ideal for treating any sort of cancer... Just find the distinct sequences and delete?
What bad news do I not (yet) know that makes it much tougher than I'm hoping?
You could, however, identify regions of DNA which make your cells more susceptible to the transformation, but CRISPR is still more unstable than the OP makes you think - in the highly debated human CRISPR experiment (using an older protocol) the scientists started with 86 embryos, of which 71 survived, of which only "a fraction" was successfully transformed, and in another fraction of that the proper target was hit but only in mosaics (with non-transformed cells remaining), in the remaining successful transformed cells the new DNA was inserted in the wrong position (potentially creating a new disease). See Carl Zimmer's take here: http://phenomena.nationalgeographic.com/2015/04/22/editing-h... There's a lot of hype which IMHO damages the technology, it's much more finicky than it's usually presented
What if someone gave Michelle Obama a very high chance of breast or ovarian cancer by giving her BRCA1 or BRCA2[0]? What if a dictator's competition got sickle-cell anemia all of a sudden, or lost their sickle-cell anemia before their trip to a malaria zone (which is also bad[1])?
[0] http://www.ncbi.nlm.nih.gov/pubmed/12677558
[1] http://www.cdc.gov/malaria/about/biology/sickle_cell.html
Consider that the only reasonable mechanisms for gene therapy on grown animals is to infect them with a virus (or something similar). If you can infect someone with a virus, they're already dead.
http://www.the-scientist.com/?articles.view/articleNo/45119/...
The article itself also absolutely stinks. First an ad hoc justification why this article is important consisting of nothing more than platitudes about science, then evocative descriptions of the weather in Santa Pola, progressively more brief descriptions of other sicentists' work followed by a long section at the end which could be a slightly edited version of the closing remarks from the Broad Institute's patent attorney in their dispute against the other CRISPR developers.
This in particular, stinks:
"The history also illustrates the growing role in biology of “hypothesis-free” discovery based on big data. The discovery of the CRISPR loci, their biological function, and the tracrRNA all emerged not from wet-bench experiments but from open-ended bioinformatic exploration of large-scale, often public, genomic datasets. “Hypothesis-driven” science of course remains essential, but the 21st century will see an increasing partnership between these two approaches."
What 'growing role'? Biology has always been largely hypothesis-free. Which hypothesis was being tested when DNA was discovered? They were trying to fit a model to the data, not testing out their ideas about how DNA should look. Penicillin was discovered by accident, where was the hypothesis there? And the importance of bioinformatic exploration? How does this: "Using his word processor, Mojica painstakingly extracted each spacer and inserted it into the BLAST program to search for similarity with any other known DNA sequence" support the importance of bioinformatics? Manually pasting 4500 sequences into BLAST? Surely this is an example of progress DESPITE any substantial bioinformatics.
I can't see why this was written except to further the agenda of the author and his institute, and to color history to support their legal claims. I can't see how this was published except through nepotism, eminence based science and the kind of shameful arrogance and entitlement that only academics have the luxury to cultivate. Maybe Lander thought he was setting the record straight, but that isn't his privilege.
I think Dr.Lander is talking about having open infrastructures and databases cataloging human genome variations; that allow everyone explore the datasets to find new patterns.
You are right searching through BLAST is easy. However, maintaining and updating entries into the NCBI BLAST databases is not. Genomes are sequenced, assembled and submitted to NCBI every day. Cataloguing the variations in different strains of the different species, and identifying real novel evolutionary variations compared to just noises to existing reference genomes is not an easy task (1000 Genomes, dbSNP, malaria genome network, etc.)
It is analogous to saying, Googling is easy but building and maintaining a search engine is not. I get what you are saying that some important discoveries are made by accidents, but the open data and intuitive platforms need to be there for the inspirational late night "Showerthought" googling-that-leads to-insight to happen in the first place.
http://www.ncbi.nlm.nih.gov/pubmed/?term=CRISPR+AND+Lander+E...
There are 5 citations. 2 are reviews. 2 are original research: one in Jan 2014, another in Nov 2015.
A similar search for Jennifer Doudna reveals 44. Feng Zhang has 43. Maybe he's been doing years of CRISPR work in secret, but I don't think my premise is 'flat out wrong' - he isn't the person to write the definitive history of CRISPR.
Second - the human genome project is surely the key example of an entire field that could not have been birthed without informatics - Celera genomics doing the shotgun assembly showed everyone how important it was beyond a doubt. This all happened in 2000. Lander is saying this isn't enough to convince us all how important bioinformatics is. In 2016, when bioinformatics is so ubiquitous, he finds the need to cite Mojica copy pasting 4500 sequences manually from microsoft word into a BLAST search in 2003 as another seminal example to make sure we all get it? This doesn't strike you as even a little disingenuous or ad hoc?
I guess we have different ideas about how problematic it is if someone tries to rewrite history while it's still happening.
2. FWIW, I think you are largely rewriting history by your comment. The dominant feeling I remember about the HGP at the time was that it wasn't worth it and we didn't learn any new bio. This feeling is still widespread... And biology has always been "hypothesis free"??? Really? Anyways, both of these points were largely ancillary imo to the early history of the field that I thought was an interesting viewpoint.
1. To me, just having your name on a paper does not equal 'doing research in the field'. In the cases of these 2 papers, he is neither the senior author, nor does the first author come from his lab. He doesn't actually cite any of his own papers in the review, after all.
2. I'm not rewriting history because I'm talking about what people think about bioinformatics in the present because of the human genome project. Lander's point is that CRISPR informs our opinion about bioinformatics today, and I'm disagreeing this is a valid point etc etc see above. I agree with you that at the time HGP happened the reaction was lukewarm. But whatever you want to say about the impact on humanity, it is indisputable that the HGP has enabled so much science that relies on bioinformatics, and elevated bioinformatics to a key part of modern life science.
Not going to get into the hypothesis free argument here, but the issue is essentially that in biology it is often impossible to specify all the necessary details of the model you are hypothesising about, which means that hypotheses have poor generalisability and reproducibility. This means that any progress we make isn't by virtue of falsifying hypotheses, but more by observation and investigation of phenomena which may have been observed by chance. Just the act of making a hypothesis does not make a science 'hypothesis driven' - those hypotheses actually need to contribute in a material fashion to the accumulation of knowledge or more importantly, insight.