Mining Drug Space

Google Wave - Mission Drug Design Federation

2009-06-07T19:02:00.017+02:00

"The Google Wave Federation Protocol may succeed email (an innovation from 1965), as the dominant form of Internet communication."
[Google Wave Federation Protocol @WP]

Google launched Google Wave. It is a mind-blowing technical platform, which might just "change" publishing, health support of patients and physicians, and any other industry, where knowledge workers are suffering from keeping found things found (KFTF) and collaboration security (see conference blogging, libel law, compliant commenting, medicine 2.0 danger). I do not think the GWave will replace collaboration tools, but it might enrich them, by facilitating information moderation and bridging (for all peers and channels).

You can check first the Google Wave presentation (almost 1.5 hours), or continue reading below.

What is the negative side?

There are two critical posts about GWave (1,2), and I comment on them here.

like instant messaging, an attention keeper. I agree, but you could say the same for phone calls, meetings, or any other some2some, or some2many communication form. So, this can be easily disarmed by providing an "I am in the work-flow or busy" button, so do not expect immediate response. Just leave me alone and do not cause a workplace distraction.
instant typing and sentence rewriting is creepy. I agree, it is like sitting in front of a washing machine, not really educational. There should be some degree of federation cache a user can trigger. Sometimes I might want to write even a full paragraph before getting viewed by others.
CCing wrong person. This is a security leakage problem, and can happen with any electronic information. It is just a gut feeling, but I would assume that GWave could even trigger a recall or hold-back at any later stage. So, it might be even safer then the actual eMail system? Anyway, I still think that all confidential data should come with a peer2peer privacy option, or any other form of encryption/decryption.
another silo of communication. A what? If anything can bridge silos, then communication. The demo has shown that GWave can integrate already many other data silos, and it provides even an open communication layer. Thumbs up for Google for making the standard open.
user-wall gardening. This is serious and opens again the question how many IDs a researcher can and should have nowadays. Why do you think are people talking about global IDs, like OpenID? Exactly, to get rid of all the many single logins. So, I rather trust Google then any other small startup for hosting my identity information. Though, any company with a lot of data has to take user data, especially patient data, serious, right last.fm?
eMail threads work, so why the Wave hype. Honestly, I had already many2many threads with overlapping replies, and believe me, eMail fragmentation is a problem. At some point I was ending up with copying eMail parts to a word processor for allowing me to follow the argumentation sub-trees. And no, a live meeting is not an option, if people are in different (overlapping) meetings, buildings, and countries.

What is in there for drug design?

The key feature is the federation of data and people sources, something like FriendFeed behind your company firewall in a secure legal environment.

"@Google, please take data security very seriously, only a few people might want that confidential data is moving to your server. We can talk about contribution tokens and embedding tools, but data should stay in the remote company silo." [personal note]

Business Intelligence and News Alerts. You might follow a series of business intelligence news, journal article news, or other news feeds. Is it not sometimes a pity, that the commenting happens all over the place? Sure, there are BackType and Disqus for a comment federation, but it would be so much nicer, if everything could be federated in one stream. The Google Wave demo showed a federation example with a blog post, so I assume this will be possible.
Dynamic Reporting. Imagine you get an updated data report for your project X. As usual, people can reply via eMail and tell you what is nice or unexpected. But, it would be so much nicer, if people could actually collaborate on the data in one workflow, instead of switching forward and backward between different data sources and communication media. This could really boost efficiency.
Data Silo Bridging. Like many other tools is the trend going towards peer and feed federation, like on FriendFeed, Yahoo pipes, or RSS readers. This are all tools with their own API. In other words, you will face the same communication problems you have with data warehouses, either you create all NxN relationships between the services, or communication will be inefficient. Why do you think have people invented BioDAS or BioMART? Exactly, for having only one service many others can talk to, just like Google Wave. This will reduce the number of communication and API bridging efforts dramatically.
Project Feedback. If I have a question within a project where the project members are not sitting in my room, my building, my campus, or even my country. What can I do? I write an eMail, and expect sometimes tree-like eMails threads, which are overlapping. If you like playing eMail puzzle, this might be fun, it is certainly not when you are having a busy work day. Then eMail fragmentation is a distraction. GWave looks promising and might just avoid this problem, totally.

Finally, I think Google Wave can certainly reduce the information fragmentation and data silo bridging problem with its new and open federation protocol.
I hope it might find its way into businesses with strict communication regulations, eg. Pharma, Medicine, or any other organization dealing with patient data. I further hope that Google takes data security and data duplication issues very seriously in their federation protocol. Google might have here an ability to tackle both things, I am looking forward to it.

Drug popularity (via Google queries) - Yet Another Long Tail (YALT)

2009-05-30T00:09:00.010+02:00

Here some drug popularity trends based on automatic Google queries. The drug names were taken from DrugBank.

The top ten are

Cholesterol
Aspirin
Ethanol
Ibuprofen
Insulin
Acetaminophen
Testosterone
Cocaine
Oxycodone
Vardenafil

References

I used a Python script for the Google data fetching. Let me know, if you need more details.
The Long Tail @Wikipedia
DrugBank: a knowledgebase for drugs, drug actions and drug targets.
Wishart DS, Knox C, Guo AC, Cheng D, Shrivastava S, Tzur D, Gautam B, Hassanali M.
Nucleic Acids Res. 2008, 36, (Database issue):D901-6.
PMID: 18048412
DrugBank: a comprehensive resource for in silico drug discovery and exploration.
Wishart DS, Knox C, Guo AC, Shrivastava S, Hassanali M, Stothard P, Chang Z, Woolsey J.
Nucleic Acids Res. 2006, 34, (Database issue):D668-72.
PMID: 16381955

Compliant commenting in science 2.0

2009-05-26T22:18:00.008+02:00

We know that scientific collaboration is important, but what is holding people back to contribute to web 2.0 communities (of people, like you!) ?
This question was raised on FriendFeed (another web 2.0 platform, a good one).

Though I answered parts of this question earlier, here a focus on the increasing number of compliance regulations, especially for people in industry (e.g. drug design). Beside, please check also the dangers in health/medicine 2.0.

Science And Compliant Commenting

View more OpenOffice presentations from Joerg kurt Wegner.

Peer recognition and building relationships - Even in drug design

2009-05-19T21:16:00.012+02:00

"Experts will want to contribute to coworkers who need them, who will hear them, who will respect them and who may even thank them. ... Rather than asking, How do we incentivise people to share their knowledge? It would be more useful to ask, How do we develop relationships across the globe that will set in motion more knowledge sharing?" [slightly generalized after Nancy Dixon]

"Today there seems to be a new distribution model that is emerging. One that is based on people’s ability to publically syndicate and distribute messages — aka content — in an open manner. This has been a part of the internet since day one — yet now its emerging in a different form — it's not pages, it's streams, its social and so its syndication." [John Borthwick]

Recently, there are many efforts ongoing to build relationships in the drug design community, and yes, the peers are recognizing and appreciating those efforts.

WikiChemists and Chemical Abstract Service (CAS) announced their collaboration. Beside, we should never forget that this is a community achievement of many people, especially the WikiChemists and everybody involved in ChemSpider (appreciated by RSC). Dear CAS, we appreciate, that you appreciate.
ChemMantis (slides) and Reflect (webcast) bridge chemical and biological data silos. This saves time and enriches information and is peer appreciated.
Biskit bridges tools and services manipulating and analyzing macromolecular structures, protein complexes, and molecular dynamics trajectories. If you ask me, any efforts making daily life easier and giving access to many tools via one interface is peer appreciated.
BioMart and DAS (Distributed Annotation System) are both systems designed for accessing many biological data sources via a unified interface. This is peer appreciated.
The Functional Genomics Experiment data model (FuGE) has been developed to facilitate convergence of data standards for high-throughput, comprehensive analyses in biology. FuGE models the components of an experimental activity that are common across different technologies, including protocols, samples and data. The community is using it and contributes to the development of convergent data standards in omics research. Clearly, peer appreciated.
When thinking about drug or side-effect re-use, a lot of information and people were involved in creating this information. So, helping patients by bringing drugs to people to drugs to people to ... is peer appreciated.
Predicting novel targets for existing drugs using side effect information

Drug design and thinking the unthinkable

2009-05-06T20:32:00.015+02:00

"The problem pharma industry faces isn’t that they didn’t see generics coming. They not only saw it miles off, they figured out early on that they needed a plan to deal with it. ... In drug design the unthinkable scenario unfolded something like this: The ability to share data and intellectual properties wouldn’t shrink, it would grow. Walled data silos would prove unpopular and inefficient. ... No one experiment is going to replace what we are now losing with designed drugs, but over time, the collection of new experiments that do work might give us the cure we need." [adapted from Clay Shirky]

We need legal support, intellectual property protection, and social legal systems, we need them more than ever ! Sure, we want to work 'openly' together, but not without payment. I think there are little intentions of patients and scientists to starve themselves to death for providing a health benefit to others. The key question is therefore how we can measure knowledge and contributions allowing us to reward the support of patients and scientists? Is there any metric or legal system we can adapt? If you have any idea, let me know !

"You can’t control what you can't measure." [Tom DeMarco]

What can we learn from

Lawrence Lessing (legal aspects of free cultures),
Andrew McAfee (Enterprise 2.0), and the
Do Tank for delivering Do Drug Development (D³) solutions

Open innovation in drug design? - I do not think so!

2009-03-30T19:30:00.017+02:00

Triggered by slides and a discussion from Cameron Neylon was I wondering, to which degree people believe in open innovation for the drug design community?

Though the principle seems valid for a few software companies, is there little support for life sciences on this topic.

"It's an area that's just not been explored much. The usual flow is very much linear. Can be developed in academia, is then licensed via tech transfer." [Deepak]

"The question is whether a business can make money by a route that doesn't involve patents (or copyright). There are examples of that, O'Reilly publishing being the most obvious but they are pretty thin." [Cameron]

In other words, if you do not need the money, do whatever you like openly.

If you need the money, e.g. for clinical trials, then please make sure staying within the same legal framework, which is most probably not open! Do not misunderstand this, I still strongly believe in (the legally encapsulated) web 2.0 !

Drug design is not like the IT industry

Some people think drug design is like chip or software design, e.g. by (over)simplifying the business models. I heard that before, and I still doubt so! Chip and software design are mostly an engineering task of known facts or an optimization of measurable parameters. In contrast, drug design uses often facts or parameters from the tip of the iceberg, but there are much more unknowns , or even things we can not look at (like *all* metabolic pathways in humans)! Beside, the exploding amount of information is making things not easier.

On the other hand, who could blame humans for oversimplifying problems ...

Unless we concentrate very hard, we often simplify a problem, because our minds routinely do so without knowing it.
Many people confuse the statement that "almost all drugs are small molecules" with "almost all small molecules are drugs". Assume that the first statement is true, that 96 percent of drugs are small molecules (not counting biotech, or nutraceutical drugs). This would mean that only about 0.01 percent of small molecules are drugs, since there are more than 24 million physically existing small molecules and only 4600 drugs, roughly one in five thousand. So the logical mistake makes you (unconsciously) overestimate the odds of a randomly drawn small molecule being a drug by more than five thousand times ! [freely adapted after The Black Swan, p 52]

References

4637 small molecule, 137 biotech, and nutraceutical drugs as taken from DrugBank 2009-03. The number of 24 millions physically existing (as published, not as existing or purchaseable) small molecules was taken from a recent CAS publication.

The future of science and scientists - walking into open innovation ?

2009-03-26T21:23:00.011+01:00

"If you just build it they (probably) won’t come...
If you don’t build it they deﬁnitely won’t come...
The community is more important than the service...
Researchers are already the tail !" [Cameron Neylon]

"An ideal collaboration market will enable just such an exchange of questions and ideas. It will bake in metrics of contribution so participants can demonstrate the impact their work is having. Contributions will be archived, timestamped, and signed, so it’s clear who said what, and when. Combined with high quality filtering and search tools, the result will be an open culture of trust which gives scientists a real incentive to outsource problems, and contribute in areas where they have a great comparative advantage. This will change science." [Michael Nielsen]

"Specialization will be one of many factors influencing creation of new types of scientists. And what are these types? Let me describe a few." [Pawel Szczesny]

Walking the walk - the practical experience of Web2 in research

View more presentations from CameronNeylon.

Etech Wilbanks via common knowledge, Lions, Tigers, Crowds

View more presentations from John Wilbanks.

Bridging chemistry islands to the web data silo

2009-03-22T16:04:00.008+01:00

Triggered by a few announcements of major publishers and web-services I could not resist to create a little chemical-web 2.0-story.

Bridging Chemistry Islands

View more presentations from Joerg kurt Wegner.

Do drug development (D3) - Innovation while information ignorance is increasing?

2009-03-18T21:27:00.011+01:00

People thought always drug design is rational. Nowadays, things might have changed (maybe) ... and technology alone is not the solution. IT might support collaboration and information management, but we need seriously (a timely) action of legal advisors on supporting democratic-action and intellectual-property-protection.

Nova Spivack has posted the interesting idea that the number of questions is faster growing than the number of answers. This forces us to ignore more facts, which leads us away from rational decision making, if we are not collaborating. I think, he has the point that in some areas we are facing bounded rationality in phases of a work process. On the other hand, I am willing to ignore information about the south of orion and rather read another drug design paper instead. The problem is that I can not read all 'relevant' articles (filter problem of information overload). Recently, I met Barry Hardy and he named it also the real-time problem of knowledge management. So, I have to filter, filter, and filter my priorities, which eliminates elements of making serendipitous discoveries, and puts me on risk of being not innovative and rational ignorant in my work.

Collaboration and data driven decisions seem to be a good way for creating innovative ideas. Though, I think we will need a lot of legal advice, open source environmentalism advice, and data security discussions before the drug design community can step into web 2.0 ! People do silly things sometimes, and even Google had already data leakage problems. Usually, are internal people the major concern for data leakage, not technology, or hackers. Data leakage is unacceptable for the drug design industry, where intellectual properties are critical for a drug design process of 10 to 15 years. Or we just need longer patent coverage. Some inventions take minutes, some decades, is it then fair that all patents have the same runtime?

Anyway, Shirky is right, we need to direct collaborative efforts. Just being open does not mean that questions are getting directed or answered (have seen this a lot). He mentions the do tank - the democratic action workshop - as solution? I certainly think that legal advisors should speed-up with closing more world2web and web2world gaps (terms from Alvis Brigis). Or is the information already faster growing than the legal procedures can protect it? To all the open data, open source, and open standard people out there ... I am still with you ... and as always, open must not mean free of charge, and I think we all need to get reimbursed or bound to licenses for avoiding that others are abusing our efforts.

The Do Tank targets the "capability gap" in practicing collaboration and forming groups among people who realize the opportunity for more collaborative decision-making in their governments, communities, businesses, or other organizations but do not have the experience, skills, models or tools to fulfill the potential.
We bring "democratic" approaches to bear on our design work -- democratic understood, not as political ideology, but as a way of life where people work together to pursue shared goals.

Just from the standpoint of a scientist with a limited reading capacity; help me in talking to people which know more about it!

Reflect - annotate protein and small molecule names in web pages

2009-03-16T19:50:00.008+01:00

The core component of Reflect is a consolidated dictionary that links names and synonyms to source data entries. It was created by merging the STRING and STITCH databases, and it currently contains over 1.5 million proteins from 373 organisms and 4.3 million small molecules..

The Reflect site annotates any web page provided, or installs a FireFox plug-in to allow annotation of any web page after a right mouse-click 'reflect this page'. Here an example for the tumor protein p53.

The reflect team of Reinhard Schneider is one of the Elsevier grand challenge finalists (sorry, for making the wrong statement before that they won this challenge, which is not known, yet).

Drug design paradigms - get involved in helping patients - haystacks, islands, and cliffs

2009-03-08T12:55:00.013+01:00

"The availability of pharmaceuticals must not be taken for granted. Pharmaceutical research requires expertise, a large commitment, a great deal of patience and the courage to take large financial risks. What counts is a sustainable framework in Germany (and other countries, too) as a location for the pharmaceutical industry. This is not a guarantee for economic success but is certainly at its foundation." [we get involved - vfa innovation]

"To bring personalized medicine forward, scientists, regulators, policy makers, and pharmaceutical company leadership will need to find ways to more successfully work together, adopt a new mindset, and take a networked approach to innovation—one in which we can successfully share pre-competitive information and common platforms such as biomarkers, genomic data, predictive toxicology, and serious adverse events information." [Paul Stoffels - Modeling change]

Why is drug design an island-bridging-task?

There are several drug design paradigms. Nowadays, it is more realistic to consider innovative drug design (report for 2008) as an island-bridging-task, then some of the earlier drug design paradigms, like serendipity and searching in a haystack.

What are cliffs we need to bridge?

patent cliffs - For having enough money for research and all regulatory steps for bringing a drug to the patient we need intellectual properties and patents for ensuring at least some rights on innovative drugs. Associated with a patent is the patent expiration. This means the network of inventors looses all rights for exclusivity and any other company can just use the final outcome of research efforts without reimbursing the inventors. Inventors (a pharma company) looses a lot of money when a drug goes off-patent. This challenge is called a patent cliff many pharmaceutical companies are facing in the next years. Anyway, we have to accept that generica companies are expanding, sometimes even by faking drug data.
activity cliffs - Do you think chemists and biologists can just make any chemical structure they want? Do you think that small drug modifications are having a small impact on the behaviour of a drug? No, this is - not - true, especially, because drugs are behaving differently in different species and because drug properties are not rarely inversely correlated.
science cliffs - Though people might assume that reviewed journal articles and patent procedures are free of errors, this is not the case. There are scientific journal publications with errors or patents of questionable value. Beside, usually are most journals biased towards positive results, only.
technological cliffs - In short, there is more chemical, biological, and literature information a single person can handle. Even worse, still, we are facing many non-connected data islands (data silos), often even without proper semantic or annotation.
communication cliffs - Some people, have a problem with collaboration, maybe just because of the NIH syndrome, or insufficient collaboration models?

What are bridges?

patent cliffs - We need a lot of contributors - in pharma - academia - and other institutions.
activity cliffs - We need support and research from many people to achieve and understand drug safety, especially for novel disease areas and targets.
science bridges - Discuss science on all levels, not only in peer-reviewed journals, join the science community on all levels, also by accepting, discussing, and avoiding reinventing negative results.
technological cliffs - We need technological, and especially social support for enriching and curating scientific information.
comunication cliffs - It was already shown that each single user contribution can enrich knowledge dramatically. For the chemistry area are the reason for missing user contributions not totally clear, but optimized technology and collaboration models seem to be very relevant.

so, get involved ... in funneling innovation ! Help eyeballing !

"Never doubt that a small group of thoughtful, committed citizens can change the world. Indeed, it's the only thing that ever has." [M. Mead]

References

D. D. Richman, D. M. Margolis, M. Delaney, W. C. Greene, D. Hazuda, R. J. Pomerantz, The Challenge of Finding a Cure for HIV Infection, Science, 2009, 323, 1304-1307. DOI: 10.1126/science.1165706
All pictures taken from Wikipedia commons (bridge, island) and compiled with Picasa.
E. Kerns, Drug-like Properties: Concepts, Structure Design and Methods: from ADME to Toxicity Optimization, 2008.
Cueva de las Manos, hands.

Nature - mission science - join the blogosphere

2009-02-28T17:33:00.011+01:00

Nature's mission statement

"FIRST, to place before the general public the grand results of Scientific Work and Scientific Discovery ; and to urge the claims of Science to a more general recognition in Education and in Daily Life." [Nature's mission statement]

"SECONDLY, to aid Scientific men themselves, by giving early information of all advances made in any branch of Natural knowledge throughout the world, and by affording them an opportunity of discussing the various Scientific questions which arise from time to time." [Nature's mission statement]

Nature's recommendation

"More researchers should engage with the blogosphere, including authors of papers in press. ... There are societal debates that have much to gain from the uncensored voices of researchers. A good blogging website consumes much of the spare time of the one or several fully committed scientists that write and moderate it. But it can make a difference to the quality and integrity of public discussion."
It's good to blog (editorial), Nature, 457, 1058, 2009. DOI 10.1038/4571058a

Chemical suppliers - CHEMCATS and the long tail

2009-02-19T20:58:00.007+01:00

If CAS is providing the head of the chemical supplier power law function, who is providing the tail?

As pointed out by Derek recently and earlier exists there a vast amount of chemical suppliers. But honestly, how many people are taking the challenge of browsing all suppliers when looking for a chemical building block ? I would guess that more than 80% of the chemists are just using SciFinder, which is based on CHEMCATS. The really funny part is that chemical suppliers have to upload their data to CAS. So, CAS is not only charging users for the look-up of chemical substances, but they are also charging suppliers for uploading their substance data, and for keeping it up-to-date.

I am wondering, could there not be other business models? Which other services could provide the long tail of chemical supplier information?

Two potential upcoming services (in the future, not populated with information right now) could be ChemVendor and ChemSpider. Both allow uploading data, and I guess it is more a rhetorical question, if an upload of chemical supplier information could be possible in the future? In contrast, both services are looking for more data upload, user, and supplier contributions.

Here are the requirements for the CHEMCATS upload, so can ChemVendor and ChemSpider allow the same? Well, lets ask them ...

Catalog name/product list name
Chemical name
Synonyms, Trade names
Catalog Number/Order Number
CAS Registry Number - Why not InChIKeys ?
Structure diagram
Grades
Purities
Quantities available
Prices

ChemSpider allows even hosting more information, e.g. Spectra, so lets see what happens in the next months.

Anyway, which other services of chemical suppliers do exist right now?

Of course, are there much more services out there, but honestly, even if I would list all of them, who would check all of them?

Life science - Are the IT guys getting it ?

2009-02-10T19:25:00.008+01:00

"My argument is this: there is so much useful information that needs to be organized in a format that is free, navigational and easy to access. One person cannot do it alone; we all need to contribute for the betterment of science, in general. ... I’ll gladly contribute! How do we get the ball rolling?" [J]

The Realm of Organic Synthesis initiated a discussion about chemistry knowledge management, which has triggered a lot of follow-up discussions.

Mitch (blog) said that chemical informatics guys do not get it:

"What J and I (Mitch) are planning is truly awesome, but it is not going to fall within the realm of the type of collaborative work that the chemical informatics people know so well. J will announce the project in the next couple of days. I think you will find it completely awesome, or simply not get it."

I am really curious to see what they are planning. I agree with Rich that many tools and databases are of questionable use for people working on a daily basis as chemists or with chemists or biologists.

Just by accident, an (industry) colleague was asking me today, if I can not recommend an open-source tool for browsing through a SDFile, which is the de-facto standard for exchanging chemical data with experimental biological activity. Honestly, my first thought was BioClipse 2beta, which I have not looked at for months. I know that people are working on it for years, but guess what, SDFile browsing was not working for the data files I tried. As quick-fix I recommended the freeware (not open source) ChemFileBrowser. Not to forget all the commercial options I told him.
Anyway, looking forward for other recommendations I am not aware of ?
I agree again with Rich that CAS number searches are important. After the earlier discussions, we know that there is little room for fair-use as brought up by Tony. Still, at the end (of what?), the CAS guys are officially supporting Wikipedia, but is ChemPedia legal with all the CAS numbers it hosts?
PubChem: Well, lets put it that way, I know it, but how useful is it in industry? Lets face it, (dirty) quantity is not everything, but quality (ChemSpider), and most important purchasebility (eMolecules). At the end of the day, we want to create compounds, not just look at them on the screen.

Anway, J and Mitch, on the one hand you say you want to collaborate on the other hand you are not interested in the collaborative work of the cheminformatics guys. Sorry, I am confused !
I really hope you do not want to create yet-another-collaboration-online-data -silo. I must agree with Deepak, that just historically, it is unlikely that a single whatever service will make it. So, we have to think about some sort of communication between services. As I commented on Egons post, I believe in the social principles, rather trust people than technology alone. Any form of knowledge is just data, and faces the risk of creating a silo: in people, in departments, and in institutions.

Let me quote you again

"much useful information that needs to be organized in a format that is free, navigational and easy to access"

What a challenge. You can have

little structured information, easy to organize, easy to navigate
much structured information, easy to organize, difficult to navigate
much unstructured information, difficult to organize, difficult to navigate, some people call this life science, and lets not even get started with the whole omics thing ;-) Well, maybe a little !

Making it even more complicated, you mentioned collaboration, so

little contribution, easy group dynamics, no technical requirements, just take a piece of paper on a round-table
lot of contribution, difficult group dynamics, massive technical challenges ranging from communication standards to performance problems.

At the end, I see some discrepancy between "lot of information" and "easy to do", so, again, curious to see what happens next. As long as people keep talking and the ball rolling, everything is possible ...

... and some IT guys are really not getting it !

Collaboration bullseye 2.0: Information overload, filter failure, and ways out

2009-01-17T18:46:00.027+01:00

"Using the Internet today is like trying to use a library where all the books have been dumped on the floor and the lights turned out." [A.J. Vendeland, 1996]

"A metaphor I use to describe the Internet and now web, how it is organized, and how useful the stuff is that one can find there. My kitchen junk drawer has lots of good useful stuff, mixed in with lots of useless stuff that I have no need for but I have kept anyway, just in case!" [J. C. Neale]

This post has two major goals. One is to disentangle the difference between information overload and filter failure in science. The second is to show a potential way-out of both problems.

This post was triggered by a controversial discussions, saying that in science information overload IS or IS NOT filter failure.

IS - Clay Shirky started the whole "It's not information overload- It's filter failure" discussion and explains why. He brings even a scientific chemistry study group example after 15' in the video.
IS NOT - Bench Marks raises three major points against this idea in the context of science. First, filtering is already done, better than ever, but still the amount of information is too high. Second, he has concerns trusting a random group of web 2.0 strangers. Third, he thinks science will suffer, if the paper reading is left to other (not cutting edge) researchers, because they might not be good enough in connecting the innovative dots.
IS - Alan Cann says that "Conventional wisdom holds that Aristotle was the last person to know everything (Neill & Ridley 1995 The Philosophy of Art: Readings Ancient and Modern. McGraw Hill). That probably wasn't true - Aristotle was a victim of Rumsfeld's Syndrome - he didn't know what he didn't know. Certainly, every scientist since Aristotle has suffered from information overload - even if they didn't know it. So if we can't know everything, what do we have to do? Filter !"
IS NOT - Bench Marks clarifies again that scientists are already filtering like crazy, but due to the exponential growth in research and journals, there is more knowledge they can assimilate. He offers also two solutions. One is to specialize more, which seems contradictory to the cross-disciplinary world out there. Second, he recommends spending more time with reading, if scientists want to work in a cutting edge area.

Before I will start with discussing actual business trends and solutions, it is important to understand that there are two major forms of information overload. I think one of them relates more to the 'classic' information overload problem (rabbit hole risk), the other more to a filtering problem (Spigot risk). Still, both things are highly related and I am not sure if any differentiation is of much use. At the end we need to tackle both problems, e.g. by collaborating.

Rabbit hole risk of information overload

"Suddenly she sees a white rabbit run past wearing a coat and carrying a watch, lamenting running late. She follows it down a rabbit hole and falls very slowly down a tunnel lined with curious objects." [Alice in Wonderland]

In short, this is the effect of highly non-linear information, e.g. encyclopedias, web pages with links, and scientific papers with references, which allow you do get easily distracted by more and more information, e.g. via links.

I stopped counting the number of deja-vu's experiences, where I was facing this effect. I started with a single scientific article, and was ending up with an even larger number of papers on my reading list afterwards. Especially, review articles have usually more references I put later on my to-do list. So, this phrase 'follow the white rabbit', is not only something for computer geeks, but also a well-known information overload risk to knowledge and science junkies. Honestly, when thinking about cutting edge scientists, then my guess is, that they are more stressed than others, just because they are always trying to stay close to the scientific oligarchy. People which have stopped trying to catch up, do not care if they do *not* read five or twenty articles, news alerts, or encyclopedia entries.

Spigot risk of information overload
The spigot risk is that the incoming scientific information flow is not only continuous, but also exponential, while filtering and reading capacity is usually not.

I would bet that anyone who tried to stay competitive has faced this problem already, especially people with the above mentioned rabbit-hole-syndrome. To get out of this dilemma, I rather trust people, than any technical solution (alone) to filter, rank, and annotate information, e.g. eMail alerts, or RSS streaming. This is also one reason why all the systems out there should allow a collaborating human factor to allow positive networking effects.

Why social software might be a solution
The evidence for this can not only be found in the wisdom of the crowds on Wikipedia, but also in several top web enterprises, just look at the top 500 web companies and how many of the top-scorers are based on social software elements.
As highlighted by Andrew McAfee (blog) is the key element to trust people at some point, which is definitely better than trusting reading or paper writing bots. He is showing how important it is to keep social networks on several levels (they are ranging from tight, weak, potential, to none). This is the major key for understanding the resulting collaboration forms based on this collaboration bullseye.

Again, why should we care?

Should we not only rely on internal knowledge and closed groups? Actually, no ! We know already the problem of bounded rational drug design. We also know that "the problem of a rational economic order is… that the knowledge… of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge" [F. Hayek].
In other words, nobody expects that a single person knows everything.We know that people are getting more specialized and that we are facing a wide range of diverse cutting-edge experts. Guess what, I think this is exactly what we need, but not on all collaboration levels. Andrew McAfee confirmed this also with the numbers of problems solved by InnoCentive. The numbers of problems solved was usually higher, if the diversity of the scientific contributors was higher. I must admit I was amazed to see that you can find a lot of chemical and life science challenges on InnoCentive (@WP).

Finally, we have to trust other people, and they are not all random, some are already in a tight or at least weak collaboration and information network with you. Sorry, Bench Marks, I would rephrase "To a hammer the world looks like nails" to

"I am a scientific nail, but I trust a hammer often. If not, I want to be close to other scientific nails. Lets make sure most nails are strengthening their networks and collaborations, not only their (closed) knowledge levels and data silos."

I would highly recommend to follow McAfee's talk after (47:51), because he addresses also the intellectual property conflict question (or how open should we be?) and mentions highly respected life science companies like Pfizer, Eli Lilly, and InnoCentive.

Beside all the things already discussed, I also mention things I find useful for finding, filtering, collecting, and ranking information.

Searching - I am using iGoogle as my personal homepage, and I have created a few gadget for helping me, but I am also using things, which were already available, e.g. a RCSB search.
Collecting - Since I am not the only one loosing time in finding relevant information, I use my very personal mixture of content management systems (CMS) like MindManager, XWiki, and a LaTeX/BibTeX solution for keeping track of my read articles.
Social networking - Well, the people which know me, are fully aware, that I like communities and that I appreciate community efforts. I think everybody, which tried to get community things going knows that it is not always easy to reach a tipping point. Some previous posts might highlight this:
- editorial, social, or both - again, I believe in Linus' law: given enough eyeballs, all bugs are shallow
- a community moves on - well, also networking costs time
- six degrees of drug design - I believe in diverse within six steps
- quality improvements are always possible
- Finally, I am working for a pharmaceutical company, so I am open, but not stupid. Anybody in this industry knows that intellectual property protection, project work, external collaborators, and any target activities are highly confidential. So, if you are working on such things, like I do, do not expose this in any form on any typical open social network tool like your blog, Facebook, Twitter, FriendFeed, Plaxo, whatever (see cons of social networking or listen to Clay Shirky's friend story). If you want to talk to people, follow the legal rules and policies !

References/Credits

via ibiblio Libraries FAQ: "Using the Internet today is like trying to use a library where all the books have been dumped on the floor and the lights turned out." A.J. Vendeland, in Review of The Software Publishers Association Europe Conference, Conference Analysis June 28, 1996.
Image: Down the rabbit hole @Wikipedia, modification was created with GIMP and is available on request.
Image: FlowingTap @Wikipedia, modification was created with GIMP and is available on request.
Image: Six degrees of separation @Wikipedia, modification was created with GIMP and is available on request.
The two forms of information overload on DigitalNatives were found via Strangers On The Net.

Antithrombotic graph visualization - Chemistry 2.0 and the ChemSpider journal

2009-01-04T15:37:00.024+01:00

"It’s not Information Overload. It’s Filter Failure. If you have the same problem for a long time, maybe it is not a problem, maybe it is a fact ... some of it is rethinking social norm ... we have continuously to improve filters, that just broke !" [Clay Shirky]

"We intend to demonstrate how modern web technologies can be used to dramatically enhance the type of information that can be communicated using web-based tools over standard online publishing approaches." [ChemSpider journal (editorial board)]

Inspired by the story about the chemical-study-group example of Clay Shirky (after minute 15), I got again reminded about the information overload problem. I agree totally with Clay that a long persisting problem, encourages a desire to get it changed. As shown by Tony from ChemSpider does the life science community have the (technical) ability to enrich chemical data.

So, are there just (social) barriers left? According to VonHippel do single users rarely show enough investment into innovations, if they think that this is only their single problem, but not the problem of a larger community (see chapter four). But wait, I was always assuming that the community is having this problem together, not only some single people living with their data silos.

Again, and in-line with Clay and VonHippel, are there good reasons to rethink social norms for improving information filter processes, e.g. the ones with buried, wrong, hidden, or shrouded data.

Now, lets have a look at an antithrombotic drug design data example, based on community web communication.
I used the Fontaine et al. factor Xa data set (EC 3.4.21.6) from cheminformatics.org and enriched it with PDB ligand structures (final structure file). Factor Xa inhibitors are inhibiting the trypsin-like serine protease, involved in the blood coagulation process.

What has this data now to do with drug design? One very reasonable question could be

Which XRay ligands are closest to the Fontaine et al. structure-activity relationship data for allowing structure-based drug design?

In short, there might be many ways to give an answer. Here, we are now interested in checking the functionality of ChemSpider and blue obelisk tools, e.g. Pybel. The idea is also to move towards web-based and scientific collaboration models, e.g. via google docs.

First, since I strongly believe in the benefits of curated data and community efforts, I was checking how much factor Xa data was already contributed and curated within ChemSpider. Honestly, I was facing a couple of challenges (or filter failures) by trying to answer this question. It is already easy to upload data to ChemSpider, but it is less efficient to find out, if those molecules are already registered there? Please mind, that especially people in industry have to protect intellectual properties and can not always upload data immediately, but maybe contribute on a technical level or after patent expiration. Based on the InChIKey (1, 2, 3) it was possible to create an InChIKey2CID script. Though, this web-based retrieval is still extremely slow and suboptimal, and certainly not usable for thousands of molecules. Nonetheless, none of the factor Xa molecules was uploaded to ChemSpider, not even the PDB ligands (see flat file result). So, still some work to do in the future ;-)

Second, for now ChemSpider supports mainly a quick similarity check on InChIKeys, neglecting stereo information. Strictly speaking is this only one very specfific way of a similarity search. In other words, users are left with the option for a PubChem search, which covers also the ChemSpider data. Maybe there exist already a web-service for returning for a single molecule query the most similar or the most five similar compounds, but I have not found out how.
At the end, it was easier to follow Noels Pybel examples (general, fp, sd) and creating a molsim2gxml script. The script will not only calculate fingerprint similarities between molecules, but also an XGML output for the data visualization.

Xa ligand similarity (top 5 compounds).
Original graph files (yEd): gml, graphml, xgml, svg.
Tulip acccepts also the GML format.

Xa ligand similarity (best compound, most similar).
Original graph files (yEd): gml, graphml, xgml, svg.
Tulip acccepts also the GML format.

Finally, we obtain two graphs showing the similarity (connectivity) of the factor Xa SAR training data (blue), test data (green), and the XRay data (red). Each node represents one molecule, each edge links either the most similar compound (lower graph), or the five most similar compounds (upper graph).
In the lower graph, we see that only two XRay structures (indicated in red) are close to the factor Xa SAR data in the training set (indicated in blue). One is a zinc cation, which is in this case not interesting for ligand design, the other is the ZEN-PDB ligand. This ligand can be found in the PDB structures 1j17, 1ql7, 1ql8, 1ql9, and 1v2k, which could be a good starting point for structure-based drug design (SBDD) and structure-property-relationship (SPR) transfer. On the other hand can we see that the overall distance between the Fontaine SAR data (blue and green) and the XRay data (red) can be easily spotted. So, be warned when trying any SPR transfer, and we have not even started talking about 3D ligands and the protein flexibility !;-)

So, yes, some of the public SAR data can be linked to Xray information and could support structure-based design. Anyway, a fast answer to the original question took some time. And could this now be uploaded to support the drug design community and avoiding this kind of filter failure problem for other community members? I am not aware that the EBI, or NCBI services allow data upload.
On the other hand, would ChemSpider and the ChemSpider journal (editorial board) benefit from additional services, like fast InChIKey checks, or similarity searches.

References

molsim2xgml python script
inchikey2cid python script
yWorks - the diagramming company
F. Fontaine, M. Pastor, I. Zamora, F. Sanz
Anchor-GRIND: Filling the Gap between Standard 3D QSAR and the GRid-INdependent Descriptors
J. Med. Chem., 2005, 48, 2687-2694. DOI 10.1021/jm049113+
N. M. O'Boyle, C. Morley and G. R. Hutchison
Pybel: a Python wrapper for the OpenBabel cheminformatics toolkit
Chem. Cent. J. 2008, 2, 5.
DOI 10.1186/1752-153X-2-5
R. Guha, M. T. Howard, G. R. Hutchison, P. Murray-Rust, H. Rzepa, C. Steinbeck, J. Wegner, E. L. Willighagen
The Blue ObeliskInteroperability in Chemical Informatics
J. Chem. Inf. Model., 2006, 46, 991–998
DOI 10.1021/ci050400b
D. Rauh, G. Klebe, M. T. Stubbs
Understanding Protein–Ligand Interactions: The Price of Protein Flexibility
Journal of Molecular Biology, 2004, 335, 1325-1341.

10.1016/j.jmb.2003.11.041

Drug designers are creating innovative ideas

2008-12-20T15:54:00.005+01:00

"The FDA will approve at least 21 first-of-a-kind drugs this year, up from 18 in 2007. And the number of black-box warnings decreased by 25 percent. The boxed warnings are the most serious a drug can carry." [salon.com]

It is well known that researchers in pharma are facing challenging times. Nonetheless, the FDA has increased the number of approved drugs last year (21 drugs compared to 18 in 2007). This in combination with the increasing innovation in pharmaceutical research gives hope for the next years.

The question for me is, can this industry still be innovative with all those layoffs, and still make it a mission possible for the next years? I know that some people think that technology can help

"The constant march of technology makes it possible for a smaller work force to do virtually the same job as a larger workforce." [10k Google layoffs]

Honestly, this sounds reasonable for the chip and software industry, but does this also work on scientific matters, especially for drug design? If we keep thinking and are ready for change, this might be possible.

Group Contribution revisited (via Fragment Based Drug Design)

2008-12-17T20:33:00.003+01:00

"What chemists really need is reliable data, and biophysical techniques" [FBDD events 2009]

(via practical fragments)

In the world of fragment-based ligand discovery, researchers hope that two fragments, when linked together, will behave at least additively: the free energies of binding for each fragment will sum together, with a multiplicative effect on affinity. In ideal cases, linked fragments will behave synergistically. But all too often, linking two fragments produces disruptive behavior, and the resulting molecule actually binds less tightly than would be predicted based on the binding energies of the individual fragments. This occurs not just when linking fragments, but in fragment merging and growing as well. Can such phenomena be modeled?
The mathematical groundwork was described more than forty years ago by Spencer Free and James Wilson at the old Smith Kline and French company, and came to be known as a Free-Wilson analysis. In a nice update of this work, Julen Oyarzabal and co-workers have applied this technique to ... a variety of different targets, including a kinase, GPCRs, ion channels, and P450s.
...
But even if synergy is elusive, the paper suggests that you’ve got a better than even shot of producing a whole that is at least equal to the sum of its parts.

For an excellent pragmatic review about FBDD see also the comments in FBDD events 2009.

Mining clinical trials

2008-12-15T20:06:00.007+01:00

"This (clinical data mining) application enables rapid extraction of information about institutions, diseases, clinical approaches, clinical trials dates, predominant cancer types in the trials, clinical opportunities and pharmaceutical market coverage." [10.1186/1745-7580-4-7]

(via open access news)

In 2007 more than seven million people died from cancer. This means there is still a lot to do for helping patients and for reducing this number. This is also the reason why each new treatment option has to follow good design criteria and sufficient testing, called clinical trial design. Typical design criteria could be:

assess the safety and effectiveness of a new medication or device on a specific kind of patient (e.g., patients who have been diagnosed with Alzheimer's disease)
assess the safety and effectiveness of a different dose of a medication than is commonly used (e.g., 10 mg dose instead of 5 mg dose)
assess the safety and effectiveness of an already marketed medication or device for a new indication, i.e. a disease for which the drug is not specifically approved
assess whether the new medication or device is more effective for the patients condition than the already used, standard medication or device ("the gold standard" or "standard therapy")
compare the effectiveness in patients with a specific disease of two or more already approved or common interventions for that disease (e.g., Device A vs. Device B, Therapy A vs. Therapy B)

Today, at the 2008-12-15, are there 20292 clinical trials with a cancer treatment published, some are recruiting, some are ongoing, and some are completed. This information can be accessed in XML format (DTD) (please read terms and conditions) for allowing an easier data mining on those large data sets. One method is for example the ability to browse through cancer vaccine trials via SpotFire, a business intelligence tool.

Other options could be to cross-link this information, e.g. by using crowd-sourcing concepts (Crowdsourcing@Wikipedia).

References

X. Cao, K. B. Maloney, V. Brusic, Data mining of cancer vaccine trials: a bird's-eye view,
Immunome Research 2008, 4. doi:10.1186/1745-7580-4-7
Linking and downloading of ClinicalTrials.gov data. Please read Terms and Conditions.

Public data sets on AWS (Amazon Web Services)

2008-12-06T17:43:00.007+01:00

"AWS hopes to provide researchers across a variety of disciplines and industries with tools to enable more innovation, more quickly" [Public data on AWS]

If you are interested in cloud computing and storage then you will be delighted that Amazon offers support to the science community. Though appreciated, do I think that the information provided for the data sets is very sparse, e.g. which conformer generation was used for the chemistry data sets? How exactly was the UGI dataset created?

Anyway, this sounds similar to the computing and storage solution of Google, called MapReduce and GFS. I do not know enough about it to understand the technical differences between the frameworks of Amazon and Google, but both are used to work with tons of data. As announced by Amazon would users (or companies) only have to pay for their user storage and computing time, but not for the public data. This is fair, but I was also wondering, if people could buy technical support as well?

BIOLOGY

Annotated Human Genome Data provided by ENSEMBL - An annotated form of the Human Genome, perfect for biological research
GenBank provided by the National Center for Biotechnology Information – COMING SOON - An annotated collection of all publicly available DNA sequences including more than 85.7B bases and 82.8M sequence records
UniGene provided by the National Center for Biotechnology Information – COMING SOON - A set of transcript sequences of well-characterized genes and hundreds of thousands of expressed sequence tags (EST)

CHEMISTRY

A 3D Version of the PubChem Library provided by Rajarshi (blog, homepage) - A 3D (single conformer) version of Pubchem, a public database of chemical structures in SD Format
UGI Virtual Conformer Library provided by Rajarshi (blog, homepage) - 80GB of data in SD format on conformers for 500,000 molecules that can be used for virtual screening
PubChem Library provided by by the National Center for Biotechnology Information – COMING SOON - A data set of information on the biological activities of small molecules

(via Egon@del.icio.us)

Python Training for Cheminformatics and Modeling

2008-11-09T16:23:00.008+01:00

"the best choice of high-level programming language for computational chemistry is Python" [A. Dalke]

Many cheminformatic/modeling toolkits provide Python interfaces, as shown in the summary below. This is a good reason to consult Andrew for a training sessions, or checking Andrew's or Noel's blogs from time to time. The various open source projects, from which a few are stalled, support Python, as well as the academic/commercial OEChem toolkit. Two very important modeling programs supporting Python are Schrödinger's modeling suite and PyMol. For a training on the last two programs, you best contact the software support teams directly.

Please let me know, if other molecular modeling/ cheminformatics/ statistical modeling programs should be mentioned here !

[source: EuroQSAR poster, A. Dalke, 2008, URL, PDF]

Support the star of life 2.0

2008-11-08T13:49:00.007+01:00

Imagine a world in which every single person on the planet is given free access to the sum of all human knowledge.
— Jimmy Wales, Founder of Wikipedia

Imagine the sum of all natures knowledge can maintain and restore health by the prevention and treatment of illness, as symbolized by the star of life.
— free interpretation of medicine and web 2.0

Data, models, or both?

2008-08-03T20:01:00.006+02:00

Correlation supersedes causation, and science can advance even without coherent models, unified theories, or really any mechanistic explanation at all. [C. Anderson, Wired, 2008-07-16]

First, I disagree with this statement. Second, thanks to bioinfoman3@delicious for sharing this information.

Honestly, I do not get it, why are people claiming that the scientific community, e.g. drug design, is similar to the chip industry or google concepts. Here, Chris Anderson claimed that data alone will replace theoretic concepts. As said by others, in the comments to his article, is data alone not information. Amund Tveit confirms this by showing that data correlations alone might be misleading, because you can find correlations in everything, even if it makes no sense. Data means any data, so if you look-up NME in google then you will not find a new molecular entity as first hit, though this it what I was expecting. So, I have serious doubts that this kind of thinking is supporting science at all, since it might lead away from the right path.

In the drug design area or other domains with chemical and biological information it is well known that the raw data web is not ready or, at best, suboptimal for chemistry. Let us assume that we have already fully equipped and finished data curation, and ontology projects, which provide information, not only data. Does this information-rich data help making new drugs without any models? I doubt so! Beside, the curation and ontology projects are far from being finished, so I doubt so heavily ! We have to accept that any data set will be always incomplete or noisy, even if clean music, book, gaming, and video data would be added. Would that support drug design or other scientific areas?
As said earlier

Just accept that we are working in an area where the chemical space is just too large for allowing us to get the complete picture of it. Thus, embrace incompleteness (bounded rational drug design). [Six Rules For Creating a Data Driven Drug Design Project, Mining Drug Space, 2007-11-02]

I completely agree, that

All models are wrong, but some are useful.. [G. E. P. Box]

By the way, if Chris Anderson is stating this at the very beginning of his article, why is he then trying to neglect the usefulness of models afterwards?

If people keep claiming that they know how to change science positively, then I am wondering, if they know that data generation itself might be a serious bottleneck. In drug design some effects, e.g. ADME and toxicity, are really hard and expensive to measure. Believe me, if the model disbelievers tell us that we have already enough data and models for understanding oral drugs in humans, then anyone in the drug design industry would be happy to not only reduce, but stop, human clinical tests and provide life saving drugs to patients immediately.

Finally, I strongly believe, that we need as much connected and information-rich data as possible (so far I align with C. Anderson), that we need good models for supporting decision making processes and saving time and environmental resources, and that we need a combination of both for innovative scientific thinking and novel developments.

Quotes from the neighbors (July 2008)

2008-07-27T21:27:00.000+02:00

==Receptors: Can't Live With 'Em, Can't Understand 'Em (In the pipeline)==

At various points in my drug discovery career, I’ve worked on G-protein-coupled receptor (GPCR) targets. Most everyone in the drug industry has at some point – a significant fraction of the known drugs work through them, even though we have a heck of a time knowing what their structures are like.
For those outside the field, GPCRs are a ubiquitous mode of signaling between the interior of a cell and what’s going on outside it, which accounts for the hundreds of different types of the things. ...

==A-β Dimers- The Long-Sought Minimal Culprit in Alzheimer's Disease? (The Curious Wavefunction)==

Following on the heels of the headline-making Nature publication that demonstrated that NSAIDs (Non-steroidal AntiInflammatory Drugs) uniquely targeted a substrate (APP) rather than an active site of the gamma-secretase complex involved in plague formation in Alzheimer's (see Discount Thoughts for a great summary) comes a paper that may turn out to be one of the important papers in the history of Alzheimer's disease (AD) research.

==NSAIDs bind to amyloid-β (Discount thoughts)==

One of the best-known features of Alzheimer's disease pathology is the formation of proteinaceous amyloid plaques in the brain. In Alzheimer's disease these plaques are primarily formed by the amyloid-β peptide (Aβ) derived from the amyloid precursor protein (APP) by the action of β- and γ-secretase. The length of the Aβ peptide varies, but the 42-residue form (Aβ42) is more likely to form plaques and fibrils.

Mining Drug Space

Google Wave - Mission Drug Design Federation

Drug popularity (via Google queries) - Yet Another Long Tail (YALT)

Compliant commenting in science 2.0

Peer recognition and building relationships - Even in drug design

Tag cloud - Mining Drug Space - May 2009

Drug design and thinking the unthinkable

Open innovation in drug design? - I do not think so!

The future of science and scientists - walking into open innovation ?

Bridging chemistry islands to the web data silo

Do drug development (D3) - Innovation while information ignorance is increasing?

Reflect - annotate protein and small molecule names in web pages

Drug design paradigms - get involved in helping patients - haystacks, islands, and cliffs

Nature - mission science - join the blogosphere

Chemical suppliers - CHEMCATS and the long tail

Life science - Are the IT guys getting it ?

Collaboration bullseye 2.0: Information overload, filter failure, and ways out

Antithrombotic graph visualization - Chemistry 2.0 and the ChemSpider journal

Drug designers are creating innovative ideas

Group Contribution revisited (via Fragment Based Drug Design)

Mining clinical trials

Public data sets on AWS (Amazon Web Services)

Python Training for Cheminformatics and Modeling

Support the star of life 2.0

Data, models, or both?

Quotes from the neighbors (July 2008)