Mining Drug Space

OpenAstexViewer - an open structural biology viewer

2008-06-19T19:58:00.006+02:00

As posted by Noel and Rich was the Java-based AstexViewer just got LGPLed and voila! Here it is, the OpenAstexViewer for structure-based drug design.

I personally was especially interested in the electron density functionality and how easy it would be creating a view on an enzymatically active-site.

As example I used Lisinopril an ACE (Angiotensin-converting enzyme) inhibitor. This class is seen as one of the success stories of rational drug design, based on the structural biology of carboxypeptidase A and medicinal chemistry.
My first thought was using Captopril, the initial drug for this class, but I could not find an electron density map and had to pick another example. Since the PDB structure (2C6N) of Lisinopril has a deposited electron density map this was now used.

After downloading the program and loading the structure and the map, I needed only a few mouse-clicks for selecting the ligand and creating a view on active site surface. Here is the result.
The drug is shown in the active site as sticks using atom coloring with having the zinc ion on the left in gray. The active site surface shows the lipophilicity of the protein residues. In orange we see the electron density map, which indicates a reasonable binding mode of this ligand.

Finally, but a little off-topic, I was also curious, if LASSO on ChemSpider identifies this molecule as an ACE inhibitor? In short, yes this target class gets one of the highest scores. Nice, having some online tools around for collecting quickly drug information.

After all of this I must say that the OpenAstexViewer is a very elegant, easy to use tool, which is extremely powerful. Has anyone tried already the scripting abilities ?

References

M. Adam, Integrating research and development: the emergence of rational drug design in the pharmaceutical industry, Stud Hist Philos Biol Biomed Sci, 2005, 36, 513-37. PDF.
PMID: 16137601
H.R. Corradi, S.L. Schwager, A.T. Nchinda, E.D. Sturrock, K.R. Acharya, Crystal structure of the N domain of human somatic angiotensin I-converting enzyme provides a structural basis for domain-specific inhibitor design, J Mol Biol, 2006, 357, 964-974. PMID: 16476442

Science - editorial, social, or both

2008-06-01T21:10:00.002+02:00

It was almost two months ago that David Bradley asked the question (via LinkedIn), if science may benefit from social software? Several people responded, and especially David Crotty started a controversial discussion based on my raised points.

First, Crotty said that 'popularity' is a terrible measure of quality. Actually, I think I agree on this! Otherwise, any popular web page would be automatically one with a higher quality. I have some software engineering background and would like to take this into account. I would argue that not only the access rate of web pages, but also the cross-linking character between them, or the number of errors per page would be good metrics. Was anyone following the dispute about the Britannica-Wikipedia comparison (comment)? Again, what should we compare? Error rate in articles, access rate, or cross-linking rate? Do we not have the same problem for journal impact factors? I think we have to accept that any model (or metric) will never represent the full truth about reality, but it might give us some additional rational for making decisions. Are any molecular modellers still with me? Yes, then I have a question. Do you believe that this metric discussion about molecular docking and scoring will ever end? If you think so, what metric or benchmark data set will lead to a fair comparison? I do not think that this is possible, but I strongly believe that the metrics and benchmark data sets around will help us to improve over time. Even using wrong metrics does sometimes help improving metrics, and therefore rationality.

Second, I said that "I cannot see how any editorial process can cope with this (information overload) problem"; he asks the question "Is he (means me with that) implying that journal editors are not 'really interested' in the papers they’re reviewing?".
Not at all! The point I tried to make is that the editorial process will assign reviews to reviewers, which will in most cases do the reviews also in their spare-time. So, at this point it does not matter, if reviewers get assigned by editors or by any other social web system. I think my point of view is not really strong on this, but I believe that missing transparency might be a major source of mistakes of any process, e.g. a peer-review process (via spreadingscience). I am not saying that transparency can avoid mistakes, but a transparent process would allow democratic voting systems, which is one type of decision making on Wikipedia. Anyway, in cases where no decision can be reached by user votings or reviewer comments, the next step will be taken by a Wikipedia mediator or an editor. The major difference is that in one case the editor has to do the assignment, while in the social software model the dedicated people would come to the editor automatically. And especially in cases where no editor is in place, users would just start editing (be bold!),till one or several users get promoted as pseudo-editor.Third, David Crotty says that "you need a better editorial oversight, not less" and he cites Rob Malda (via Wired) "When you're building a system like this (means Slashdot) you're balancing the wisdom of the crowds versus the tyranny of the mob. Sometimes a crowd is really smart, but some things don't work so well by committee. Crowds work when you have a tightly knit group of people with similar interests ..."
Well, again I agree that you need rather more information than less. Though, I would like to challenge this a little. Where is the editor getting his information from? I would assume that he will use at least some social web services or (social) cross-links in journal publications for finding other experts.

I was never a friend of black and white views. I would conclude that

any editorial process without some social software will run into an information overload problem, and you should not do all of this alone, but share workload.
any social software process without editors runs into an information overload problem, and how can you discriminate information from noise? You need experts, editors, and some people call this ontology knowledge, or the semantic web.
a combination of both, editorial expert knowledge in combination with social software is promising for creating massive knowledge networks, e.g. WikiProteins. Finally, any technological process will fail, if the people behind it, are not dedicated enough, run into workload problems, or are hold back by unclear intellectual property situations. So, whatever you do, do it in a team and make it a transparent process. If required, put some license, copyright, intellectual property, or whatever restrictions on it, so make steps towards users and do not wait till users will adapt to your user scenario. In all cases, avoid destroying information and allow a cross-linking between information.
As already said on ChemSpider, "get the hell organized..." and talk to each other.

Finally, David Crotty and David Bradley, thanks for discussing this in an open-minded setup and via a social software system. I am looking forward to more news, questions, answers and controversial discussions ... at the end it is all about learning and helping people, right?

WikiProteins - May the community be with you

2008-06-01T12:32:00.002+02:00

"We call on a 'million minds' to annotate a 'million concepts' and to collect facts from the literature with the reward of collaborative knowledge discovery. The system is available for beta testing at wikiprofessional.org" [DOI 10.1186/gb-2008-9-5-r89]

The author list is impressive (e.g. Jimmy, Prof. Ashburner), as well is the number of communities and organizations (PubMed, Google, Yahoo, UniProt). Of course is this not a guarantee for success, but at least is it interesting that those people have started talking to each other. I guess they all want to tacke a challenging scientific problem, which is creating knowledge out of information noise. Very impressive, indeed!

I created an account and the system looks very beta at the moment. This early release strategy is very normal, at least for Jimmy, which follows the release soon, release often paradigm of open source communities. When Jimmy released Wikia, he got also some negative feedback, because some people thought it was too early releasing the system. Anyway, this has not stopped us from founding the Life Science Group on Wikia, though I admit that not too many people have contributed so far. I hope that the WikiProfessional system will be able collecting enough critical brain mass for getting a good head-start.

"The first release of WikiProteins contains an embryonic version of what is intended to be developed into a fully functional WikiProfessionals in 2008 and beyond. Users are able to review their pre-constructed (recent) publication list and create their Knowlet before registration. With an increasing number of authors having curated their own Knowlet(s) in the system, creating communities of expertise and indicating their availability for comments and peer review, instant messaging and web conferencing will become available in the system." [DOI 10.1186/gb-2008-9-5-r89]

The system highlights three major use cases:

Community annotation: The basic principle of community annotation is that computers and experts interact in an iterative process of mining and curation.
Knowledge browsing: This will allow users browsing through the concept space of interesting relationships.
Collaborative knowledge discovery (example): When the connections in the concept space around antimalarials and tegafur are explored further, it becomes immediately obvious how logical it would be to reason that tegafur might indeed inhibit growth of malaria parasites, at least in vitro.

May the community be with you ...

Reference

Article (macmwdomchmpplbmbwmmrbb08)
Mons, B.; Ashburner, M.; Chichester, C.; van Mulligen, E.; Weeber, M.; den Dunnen, J.; van Ommen, G.; Musen, M.; Cockerill, M.; Hermjakob, H.; Mons, A.; Packer, A.; Pacheco, R.; Lewis, S.; Berkeley, A.; Melton, W.; Barris, N.; Wales, J.; Meijssen, G.; Moeller, E.; Roes, P.; Borner, K. & Bairoch, A.
Calling on a million minds for community annotation in WikiProteins
Genome Biol, 2008, 9, R89. DOI 10.1186/gb-2008-9-5-r89. PMID 18507872
Six degrees of drug design, Mining Drug Space, 2007-08-17.
Six sigma in drug design, Mining Drug Space, 2007-09-15.

Lasso, docking, and protonation

2008-05-17T22:45:00.000+02:00

The SimBioSys blog posted several literature references and discussions about protonation states

CAS numbers are not public domain, are they?

2008-03-08T10:00:00.011+01:00

"Work created before the existence of copyright and patent laws also form part of the public domain. The Bible and the inventions of Archimedes are in the public domain. However, copyright may exist in translations or new formulations of this work." [Wikipedia]

As posted by Tony is the Chemical Abstract Service (CAS) discouraging using their CAS services for assigning correct CAS numbers to structures for any third party database. Wikipedia is a source of structures, which is public domain due to its GNU FDL. Still, this does not imply that any translation of structures, e.g. CAS numbers, are in the public domain, too. Honestly, this raises a serious problem for curating CAS numbers on Wikipedia and this raises indeed the question, if they should not be dropped from Wikipedia, and any other information source, at all? Is it not better having no information, than having wrong information?

A CAS number is for me only one certain translation of a chemical structure. In this case, the only source and creator for CAS numbers is the American Chemical Society. CAS claims that their services can not be used for curating other data sources. Does this also mean that people can not use CAS numbers from publications?

"The public domain can also be defined in contrast to trademarks. Names, logos, and other identifying marks used in commerce can be restricted as proprietary trademarks for a single business to use. Trademarks can be maintained indefinitely, but they can also lapse through disuse, negligence, or widespread misuse, and enter the public domain. It is possible, however, for a lapsed trademark to become proprietary again, leaving the public domain." [Wikipedia]

And does this also mean that scientists are basically not allowed publishing CAS numbers and structures in scientific publications? How can CAS numbers then be used at all, if we can not store this information?

Many questions, and who will answer them? And, if they got answered what is the way forward for getting curated structures within the public domain? I would say (again, this time in nicer words): 'get organized scientists worldwide!' If CAS can do it, we can do it? It may take longer to get the party started, but if we do not start it will never happen.

"An ideal collaborative resource would be designed for large-scale data mining, contain curated historical data, and have data standards and deposition tools that could constantly bring in data from the published literature. ... In other words, the party might take longer to get started than hoped for, but it should be worth the wait." [M. Baker, DOI 10.1038/nrd2148]

Yet Another Life Science Informatics (YA-LSI) networking portal

2008-02-25T19:49:00.002+01:00

There are yet another two new social Life Science Informatics (LSI) networking portals. In contrast to the already available LSI networking services are those technologically very mature and offer varieties of contribution and sharing possibilities for users.

Wikia: Life Science Informatics
Plaxo: Life Science Informatics

Both networking groups are looking for a stunning LSI logo, so please feel free making any suggestions.

Social search in drug design and what is Wikia

2008-02-17T20:07:00.004+01:00

Social search engines, another kind of social software, are trying to include user knowledge for improving search results in the future. One example is Wikia with an announcement of Jimmy Wales

Search is part of the fundamental infrastructure of the Internet. And we are making it open source.

In contrast to the problem described by randfish might this improve search quality and reduce search engine tweaking requirements by users. Sometimes users have e.g. to multiply search terms or to find other tweaks, for getting the search results you want. Early initiatives like the the open directory project DMOZ (@Wikipedia) are using already user organized classification schemes for web links. SWiK organizes open source project information in a Wiki-like manner.I am still a little bit puzzled to understand what the actual difference to social bookmarking systems is. Are social bookmarking services not already ranking the relevance of web links?Lets do a test on some drug design search terms. I added the tags E(experts/users), R(anking), M(ini information page). Here a '+' indicates that this feature is available a '-' means that this feature is disabled at the moment (e.g. in Wikia) or not available (yet).

Term	Wikia (+-E,+-R)	del.ico.us (+E,-R)	Digg (+E,+R)	Technorati (+E,+R)	CiteULike (+E,+R)
Data mining	36247-M	yes	615	14630	496
Drug design	34964-M	yes	285	14966	493
Bioinformatics	16937+M	yes	75	4698	496
Structural biology	7925-M	yes	60	1758	499
Organic synthesis	4403-M	yes	15	1979	498
Medicinal chemistry	3216-M	yes	45	1686	498
Kinase	2843-E-M	yes	30	1248	498
Molecular modelling	2261-M	yes	30	208	491
Cheminformatics	460+M	yes	1	147	60
GPCR	252-E-M	yes	1	121	497

Wikia is still an alpha version and user ranking of links seems to be somehow available, but does not work (yet). Though, I like the general appearance and hope to see there some new features soon. Using del.ico.us as search engine is just useless in my opinion, because I have not found any way ranking the results. Digg might be in theory useful, but lacks scientific content. Technorati is rather a news alerting service than a search engine. CiteULlike points towards scientific articles, but why are all categories delivering less than 500 hits here?

Pipeline pilot: Student edition

2008-02-11T00:26:00.000+01:00

As posted by Richard Compton on CCL.net provides SciTegic (part of Accelrys) a student version of Pipeline Pilot (PLP), which is free for academics ! I seriously like the thinking of those SciTegic guys, they are sharp, critical, and very trendy. This is a good starting point for having a hell of a discussion about science. Smart move, which is following the integration principle !

This shows another fantastic example of a smart licensing strategy, which will probably always be critically discussed in the drug design area? Just recently there was a discussion on CCL and I commented on it. Anyway, this particular case will support science and allows scientists integrating their tools on a global level. Furthermore it allows sharing tools with industrial partner without forcing them creating an additional interface layer.

Beside, is this really kicking-ass news for everybody involved in KNIME, since people in academia can now support PLP directly (if they want). KNIME provides already lots of toolboxes, but compared to PLP this number is still small. Just check them out and especially the high number of different ways of integrating data, processing data, and integrating external tools with different languages.

Chemical webs, nets, and spiders

2008-02-04T23:38:00.000+01:00

I love it, when people come together and try to understand why particular problems exist. In September 2007 we tried contacting several people for solving the problem of having proper chemical indexing or/and search services within Wikipedia. Some people replied, but there was still this strange feeling that there is not only one responsible person, but many. How can you adress a problem if nobody is really in charge and how to take decisions in such situations? Well, maybe spreading the news had some effect, because some things happened since then.

Tony (aka ChemSpiderMan) had a discussion with Walkerma from the chemistry group on Wikipedia. It turned out that both share some common interests, which is providing curated chemical data. This is a noble goal and as summarized not an easy one, especially not because there exist already some chemistry boxes at Wikipedia.
Tony and some other people from the advisory board are showing interests in creating WiChempedia, which has also one major goal, "what can be done to facilitate delivery of quality information to the chemistry community?".
Walkerma initiated a regular discussion on Wikipedia (via IRC) for helping to understand the technical and scientific issues for incorrect chemical information on Wikipedia. Though I must say: I do not like the meeting time, which is during business hours and not practical for me.
Tony is not only moving on with extending his net of dedicated persons, but also attracting new chemical sources, and spreading publications. Looks like he is a professional advocat and knows what he is doing.
Beside all of this has Chemspider itself updated and extended his structure deposition process (PDF slides), which allow people curating data there directly.

I am usually not easily impressed, but I must say that what was done so far is a great piece of work. I do not know what others think about it, but knowing the pros and cons of social networks, it is fun to see that ChemSpider is moving with its net, and not being hold back by it !

ChemSpider search on iGoogle

2008-01-06T01:09:00.000+01:00

Extending the PubChem and eMolecules gadgets, I created a simple interface for a basic ChemSpider interface for iGoogle.

I noticed a few issues and hope that those might get fixed

The InChiKey search on ChemSpider does not support searching only for the structure without stereochemical information, which would be just the first part of the InChIKey: structure-stereoinfo. So, I hope his could get fixed.
I tried finding more search options, e.g. defining substructure or exact searches, but could not find them in the ChemSpider tutorial. If there is more detailed information around I would appreciate getting some hints.

And just for clarity, here is a list of the other gadgets I have created

Boutique drugs: Splitting mass market and drug targets

2008-01-05T22:15:00.000+01:00

"Recently, the term boutique has started being applied to normally-mass-market items that are either niche or produced in intentionally small numbers at very high prices. For example, before the release of the Wii, a Time Magazine article suggested that Nintendo could become a boutique video-game company, producing games for niche audiences, rather than trying to compete directly with Microsoft and Sony [1]" [Boutique@WP]

"articles intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man or other animals and articles (other than food) intended to affect the structure or any function of the body of man or other animals." [FDA definition]

As found via Bio-IT/drug discovery expect analysts in the future more boutique drugs. Two main causes are mentioned.

First, the regulatory hurdles for getting a drug on the market have dramatically hardened since 2004. Furthermore has the number of black-box warnings (FDA guidance) at least doubled since that time.
Second, is there still a dwindling pipeline for new medications.

And I think Garnier and Brozak have the points when saying that it is difficult to react in a short time to many new drug projects/targets:

"The hurdle has been raised, there's no question about it! What is unclear is exactly how we should modify our drug development plans to meet this higher hurdle." [J. P. Garnier]
"No pharmaceutical executive has ever lost his job for saying no to a new drug project. And no FDA employee has ever lost his job for saying no to a new drug application." [S. Brozak]

Anyway, with a recent sad history for scientists it is to expect that drug design faces additional challenges in the next years, which are: changing drug design pradigms, avoiding paralysis.

I think it is possible, but I must say that some ideas are coming from people knowing the business, but not the science! On the other hand do "scientists, taken as a class, not always turn into the best managers" (via in the pipeline). Especially in this setup, changing seems difficult, because many are used to an established "blockbuster" drug design setup, which has worked for decades. Now, they have to define a way-out facing an innovaton funnel problem as well. If some of us scientists survive the next years, we can tell our grandchilds how some people and companies were able creating disruptive innovation helping patients in the upcoming decades.

References

Flat, round pills @WP
No smoking sign @WP

A community moves on, merry XMas and fantastic New Year 2008

2007-12-22T19:30:00.000+01:00

As posted by Deepak and Egon have several chem-/bio-informatics community members reached another milestone by publishing about userscripts in life sciences. In short, I am a believer in this technology. When reading, I got a little bit disappointed, because I got left with the feeling that I start loosing the touch to the community, though I try to follow and contribute as much as possible.

Loosing grip became also clear when I read the posts of Ola and Egon. I know that the community needs a self-selecting oligarchy for moving forward. You guys meet regularly in workshops and user group meetings. Some people might not have the time for this. In other words, you are part of the hard-core development team. This means, if you are not explaining the problems in more details and post more information somewhere, you might exclude some people, which do not have the time anymore digging trough the code themself.

As said earlier, I am working now in drug design with all the good and bad things about it (just read a few of Derek's posts and comments). This means my focus lies now on lead optimization, reading a lot, literature management, and data mining/graph mining, and only partially source code (if I can not call any support). This just means that I am very busy with many things, some of them I will and can not talk about, and this is good for my company community. As said before, data is not closed in general, but people have to accept some policies and confidentiality agreements. If you really want it, you have to talk to the right people and you need to have some very convincing arguments why you need this data.

While talking about policies, congratulations to Antony driving collaborations on several levels.

If there is something which helps people like me, for finding relevant information quickly, then that are two things:

userscripts, or in general social software, helping me finding information quickly
open standard databases, which allow people to contribute on several levels. Please note that I still think that open standards are more important than open access. 'Open' is for people in industry just a question of pricing, though I must say that some pricing strategies are just ridiculous and couterproductive for the work of scientists.

References

E. L. Willighagen, N. M. O'Boyle, H. Gopalakrishnan, D. Jiao, R. Guha, C. Steinbeck, D. J. Wild, Userscripts for the Life Sciences, Bioinformatics, 2007, 8, 487.
DOI 10.1186/1471-2105-8-487

Too many candles - World AIDS day 2007

2007-12-02T22:27:00.000+01:00

"Although treatments for AIDS and HIV exist to decelerate the virus' progression, there is currently no known cure." [AIDS@Wikipedia]

Help fighting AIDS !

source: amfAR

MetWare: metabolomics database

2007-11-25T18:11:00.000+01:00

via chem-bla-ics

The Applied Bioinformatics at PRI group in Wageningen and the group of Steffen Neumann in Halle have started the MetWare project on Sourceforge to develop opensource databases for metabolomics data.

The databases design will be based on and ideally compatible with proposed standards like ArMet (DOI:10.1038/nbt1041) and those recently written up by the Metabolomics Standards Initiative (see the issue around DOI:10.1007/s11306-007-0070-6).

One important design goal is that the project will use BioMart, which will allow easy integration of the database content in data analysis programs like Taverna and R using the biomaRt package (see DOI:10.1093/bioinformatics/bti525).

Once upon a cyclopentyl a cyclohexyl followed, and they lived not happily ever after ...

2007-11-08T23:14:00.000+01:00

via The Curious Wavefunction

One common strategy used by medicinal chemists for enhancing potency is to add a methylene carbon in a ring that's fitting into a lipophilic pocket. A methylene group usually contributes about 4 kJ/mol of affinity....or maybe not.

Gerhard Klebe's group at Marburg are studying inhibitors of thrombin in which they replace a cyclopentyl group by a cyclohexyl group. To their surprise they observe (DOI: 10.1002/anie.200701169) no change in binding affinity. Crystallography indicates no electron density for the part occupied by the cyclohexyl group but robust density for the cyclopentyl. The authors conclude that while the cyclohexyl binding is enthalpically unfavourable, entropically the six-membered ring can flip and twist and dance, which is favourable, and also does not provide much crystal density because of its flexibility. So unfavourable enthalpy is balanced by favourable entropy. MD simulations support this contention.
"What lesson can be learned from this example? Usually ligands are optimized in congeneric series. The addition of functional groups is expected to enhance binding; for example, the change from a five- to a six-membered ring should augment binding affinity by approximately 3–4 kJ/mol. However, even very similar ligands can exhibit very different binding properties that destroy a simple structure–activity relationship"
Also, the gain in binding affinity from a methylene depends on the system. Model systems based on octanol-water partitioning offer a value of about 0.7 kcal/mol. But of course octanol unlike a protein does not have a problem reorienting itself around the substrate. Thus, the value can be less for a protein. On the other hand, in cases where an extra methylene or methyl just about fills the remaining space in a pocket and packs tightly can provide 5 times as much binding affinity. This is true for example in case of tRNA synthatases (I got this from Alan Fersht's excellent book) As usual, exceptions abound to the generalisations, and Klebe's study provides another interesting example.

References

Cyclohexyl C1CCCCC1
Cyclopentyl C1CCCC1
C. Gerlach, M. Smolinski, H. Steuber, C. A. Sotriffer, A. Heine, D. G. Hangauer, G. Klebe, Thermodynamic Inhibition Profile of a Cyclopentyl and a Cyclohexyl Derivative towards Thrombin: The Same but for Different Reasons, Angewandte Chemie International Edition, 46,8511-8514, 2007. DOI 10.1002/anie.200701169

Andy Grove: When silicon indoctrinates carbon

2007-11-07T22:11:00.000+01:00

Andy Grove is right and we, the drug design community, can learn something from the chip industry. Derek, can you not see it, too? This strong relationship between those two families ...

"If you want to understand why something happens in business, study the disk drive industry. Those companies are the closest things to fruit flies that the business world will ever see. Drug design is a process between 9 to 15 years! So, which object to study lies in-between a fruit fly and a hard-disk? A high-throughput screening, a biological assay or an 'in silico' 3D/2D/xD model of a drug?" [Mining Drug Space]

Commenting on: Grim days for pharma

2007-11-03T02:35:00.000+01:00

via business|bytes|genes|molecules

Bloomberg says that drug sales are forecast to grow at their slowest ever rate in 44 years.

The WSJ reminds us that FDA approvals for NMEs are slow, and on a downward curve.

Perhaps it’s time for the pharma industry to really think about Pharma Futorology

Some of the consequences have been announced, and I think everybody hopes that those things will not happen every single year!

Can innovation help us for making drug design a mission possible? If yes, what are potential key factors for success?

Aligning scientists and goals

Putting scientists in the driver's seat
Goal congruence
Positive self-image
Fair evaluation system for high-performing scientist

Industral cultures that foster innovation

Take calculated risks
Openness and transparency
Adapt to constant changes
Facilitate effective collaboration

References

X. Zhong, G. B. Moseley, Mission possible: managing innovation in drug discovery, Nature Biotechnology, 2007, 25, 945-946. DOI 10.1038/nbt0807-945

Quotes from the neighbors (November 2007)

2007-11-02T23:07:00.000+01:00

==The value of feature extraction==

I am a strong believer in the power of machine readable data, and the potential power it can provide. Whether it be a system like Freebase that allows you to add structure to data, or the kinds of entity extraction and data contexts that Jon Udell and Jeff Jonas talk about, making your data more “intelligent” is something that we should strive for, whether it be in the world of business, or the world of life science (or as Andrew Walkingshaw might point out, materials science). My old colleagues at Scitegic have a motto - “Ask more of your data”. That has always resonated with me. So lets take yet another look at how to make your data smarter, especially when the semantic web seems to be inching towards more mainstream acceptance.

==Compiling the InChI Toolkit to Pure Java Bytecode with NestedVM==

Many cheminformatics libraries are written in C and C++; being able to reliably and automatically port them to Java could potentially save a great deal of effort.

==Come On. Improve, Already==

And man, have there ever been layoffs in the industry this year. There’s a list of the larger ones over at FierceBiotech, and it does not make for cheerful reading. In January, Pfizer announces 10,000 job cuts and closes their Ann Arbor site. That same month, Bayer-Schering closes the doors on research buildings in Connecticut and California (layoffs which were announced in 2006 and are thus not on the Fierce list). Bayer-Schering, who really should have run that B-S initial thing past a couple of native English speakers, announces 6,100 more job cuts in March. In July, AstraZeneca doubles down on its earlier layoff announcement and says that 7,600 jobs will disappear, and J&J announces a 4% reduction in its workforce (5,000 jobs). Then in August, Amgen cut over 2,000 jobs of its own.

Six Rules For Creating a Data Driven Drug Design Project

2007-11-02T19:14:00.000+01:00

"The problem is that we live in the most perfect imperfect medium in the world: the Web.
For now it is impossible to collect data perfectly. It is ugly, it is dirty, it is incomplete and no matter how hard you try it is not going to get perfect.
Yet we can’t resist." [Occam's razor]

Hahaha ... what a great statement of Avinash. Lets translate his statement about "Six Rules For Creating A Data Driven Boss!" into a drug design setting for getting "Six Rules For Creating a Data Driven Drug Design Project!".

# 1: Get Over Yourself

"In the past, many experimentalists shared a similar sentiment regarding of the usefulness of theoretical and computational methods to their drug discovery problems, namely: You're always explaining to me after the fact why a molecule i've already discovered is active, but you're never able to tell me ahead of time which molecules I should be investigating." [VanDrie]

If you have a better reasoning about drug design then random guessing then you should really get over yourself and give your opinion. You are working in a team, so appreciate challenging questions and feel free to challenge others as well. Otherwise, everybody can still hope for serendipity. But ...

"Hope is not a strategy" [H. R. Clinton]

# 2: Embrace Incompleteness

"I have heard that if the 'drug universe' included all the possible drug--like molecules, its size would be 10^62 molecules (other authors have proposed 10^200). Even if half of the currently known chemicals were drugs (a very optimistic assumption), this would produce 10^7 molecules. Therefore, at most, only 10^-53% of the possible drug--like molecules are known." [R. Lahana]

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).

Sometimes even rational ignorance might help a project, but you have to know how much effort it would be creating a complete data set.

# 3: Give 10% Extra
If you are asked to deliver results, then try to give 10% more then requested and show that you are rather a drug data ninja than a reporting squirrel. Information rules ... prepare yourself for answering questions and for having supporting information available! You have not to be a dictionary, but having 10% extra information is not too difficult.

"Make a conscious choice what job you want to be in: Reporting Squirrel or Analysis Ninja ?" [Occam's razor]

# 4: Become A Marketer

"If you want to change your boss and your company then you’ll have to become a Marketer, someone with an understanding marketing principles, someone who can be a customer advocate / champion, someone who can evangelize the purpose of data in creating customer centric decisions." [Occam's razor]

Everbody in drug design knows that 'real' data is the key for developing 'real' drugs. It costs many efforts and many people creating it. Drug design is not clicking buttons and counting bits and bytes. This is a big difference to web analytics. In drug design the data creation is an active process of the service provider, while in web analytics the data will be produced by any visitor to a web page (service provider).

Creating a new data point needs a team of passionated persons, so appreciate any single data point.

After the data is available, everybody in the team will be really excited to know if this data point will be the ultimative blockbuster.

Make clear, that data might help reducing risks. And even more important, do not sell things the data does not provide! Allow the data to influence a project, but avoid that overinterpreted data is misleading a project. Following marketing strategies might be one way of ranking data.

# 5: Business In The Service Of Data. Not.

"It is important, nay critical, to constantly check yourself and ask if the business is really serving the needs of data or vice a versa." [Occam's razor]

As said before is the data generation in drug design a time consuming and costly business. This means it is crucial to check if you can really learn from the data.

Ask yourself: Are we learning from the data?

If the answer is no, then something is wrong. Why are you producing data which costs time and money? Is the team not wasting their efforts?

Do not spend all your time in analyzing the data. Spend time in talking to the team for producing interesting data in the future.

Where do you want to go ? Go ahead! Avoid recursive data generation and conclusion cycles, which can keep you busy a while, but certainly not employed for the next years.

# 6: Adopt A New Mindset, Expand Horizons: Drug Design 2.0.
In drug design it does not count how many drugs you produce, it only counts how good they are.

Business intelligence (BI) and business performance management (BPM) can optimize the process. So what? If you are in a deadloop, it does not matter if you are in a normal or an optimized deadloop for data generation and data analysis? Get out of it!

Check what the competitors are doing! And if you are lucky, you can find this information in less than six steps ;-)

Resources

Thinking Man @Wikipedia
J. H. VanDrie, 3D Database Searching in Drug Discovery, Network Science, 1995. URL
Infinity @Wikipedia
Squirrel @Wikipedia

Uniprot: Over a billion protein database entries

2007-10-25T01:39:00.000+02:00

As posted by Eric Jain (via Egon) are now more than a billion RDF available.

The latest release of the UniProt protein database contains just over a
billion triples*! PRESS RELEASE :-)
The data is all available via the (Semantic or otherwise) Web:
http://beta.uniprot.org/
...or can be bulk-downloaded from:
ftp://ftp.uniprot.org/
* Counting some reification statements, and assuming no overlap between
"named graphs".
P.S. This should be the last you'll hear from me on this topic -- I'm off
to new adventures...

Furthermore is it notable that

The UniProt Knowledgebase (UniProtKB) and NCBI Reference Sequence (RefSeq) in conjunction with Entrez Gene are widely used resources for protein information, including function, classification and cross-references. Prior efforts in ID mapping (iProClass ID mapping table: http://pir.uniprot.org/search/idmapping) and recent collaborative efforts by UniProt and NCBI has resulted in the mapping of UniProtKB and RefSeq/Entrez Gene (GeneID) and incorporation of reciprocal cross-references in these databases. These cross-references will provide up-to-date links between the UniProt (http://www.uniprot.org) and NCBI databases (http://www.ncbi.nlm.nih.gov/RefSeq). Improving these mappings and increasing their coverage is an ongoing process.

Finally, is there a nice wink tutorial for diabetes around, which can be started here.

SuperTarget, Matador, ChemSpider: When data curation becomes a hype

2007-10-24T01:50:00.000+02:00

Although the first version of SuperTarget with all the searchand discovery tools around drug-target relations is alreadyan extensive resource for both large-scale research and in depthanalysis, the captured knowledge is still far from completeand we would like to invite the community to help in increasingquality and quantity of the records. SuperTarget offers an optionto upload and incorporate drug-target relations into a workingqueue. Uploaded entries will be reviewed and approved by anannotation team comprised of graduated scientists. Both SuperTargetand Matador can be used as knowledge sources, discovery toolsor training sets for various applications in chemical biologyand elsewhere. [DOI 10.1093/nar/gkm862]

Dear community, as already highlighted earlier is it a logical step that public data curation services like SuperTarget^, Matador, and ChemSpider become more popular. Lets see how far those services might go, but they should make things more efficient and help bringing light into the chaos. And yes, we are still living in a chaotic (data) world and every help counts.

For more access details for the databases contact the developers and authors. Here is some basic information.

SuperTarget and Matador are available via their web sites, http://insilico.charite.de/supertargetand http://matador.embl.de. They can be obtained via a CreativeCommons Attribution-Noncommercial-Share Alike 3.0 License. [DOI 10.1093/nar/gkm862]

You have limited rights in this regard. You can only assemble a database of 5000 structures or less, and their associated properties, from our database without our permission. You can download up to 1000 structures per day from the website. Please contact us at feedbackATchemspiderDOTcom to request an extension outside this constraint. We are willing to provide the ENTIRE database of ChemSpider structures at your request - the file will consist of InChI Strings, InChIKeys and ChemSpider IDs. These constraints are under regular review so please feel free to engage us in conversation. [ChemSpider FAQ]

References

Günther S, Kuhn M, Dunkel M, Campillos M, Senger C, Petsalaki E, Ahmed J, Urdiales EG, Gewiess A, Jensen LJ, Schneider R, Skoblo R, Russell RB, Bourne PE, Bork P, Preissner R.
SuperTarget and Matador: resources for exploring drug-target relationships,
Nucleic Acids Res., 2007, ASAP alert.
DOI 10.1093/nar/gkm862
PMID 17942422