About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Ignoring Patents? | Main | Traffic Record »

October 1, 2009

How To Get Useful Data Out of a Drug Patent

Email This Entry

Posted by Derek

Yesterday's post on citing the patent literature prompts today's. I realize that if you're not used to reading the things, patents can be rather odd and daunting. But there are some rules to follow to let you get useful information out of them. That depends on what you're looking for, though.

If you're searching for preparations of particular chemical compounds, there's an awful lot of that available, although sometimes it's not easy to extract. What you'll probably want to do is skip directly to the experimental section. Ignore the description; ignore the claims - if you want to find out how to make Compound X, you won't find the details you're looking for there.

Problem is, the experimental section will not necessarily be laid out in a user-friendly manner. It doesn't have to be, and in fact, it's sometimes deliberately convoluted. The first thing to make sure of is that you're looking at a real experimental procedure. If it's written in the present tense, and/or with vague details ("A compound of Formula 1 is combined with a base, preferably a tertiary amine, and is heated to between room temperature and 100 degrees C. . ."), then you'd better keep turning pages. This is not what you're looking for. This is a generic or prophetic example, and doesn't necessarily have anything to back it up. Keep going until you hit real amounts of real compounds, and real spectral data. Patent applications, ones that expect to stand up, anyway, have to exemplify the compounds that they're claiming, and that takes real data.

If the patent is written with structures over each each experimental procedure, then your job is a lot easier. Often the preps will be in sequence, one intermediate leading to another, although they don't have to be. Keep paging through until you've found the compound you're after. If the compounds are named in the procedures, a text search for some part of the particular name can save you a lot of time, as can a search for the name of some reagent that you may have seen that the inventors used (from a SciFinder search, for example).

Keep in mind that the particular compound you're looking for may appear as part of a table. Often patents will be written with a detailed general procedure for a particular example, and then there will be a line like "Following this same protocol, the following examples were also prepared. . .", and a table. I know that you'd rather have a detailed individual prep (so would I), but these are still generally pretty reliable. You find this sort of thing for lists of amides or Suzuki coupling products, so it's not too bad.

The amount of spectral data you'll get will vary. Actually, the rule for a patent application is "The more, the better", so a good one will include NMR data, and whatever else they can think of. The minimum is the LC/MS retention time and a few ions, which isn't all that helpful, but satisfies the legal requirements, barely. (Taiwan and a couple of other countries will often balk at that sort of thing, but that doesn't become an issue for a few years).

If you're looking for evidence of new biology or mechanisms of action, that will be a bit trickier. This will show up in the claims, but claim can be structured in a rather Byzantine manner. As with the chemistry, though, if it's something really important, there will be data to back it up. Gels, sequences, purification procedures - all that has to be in there if the company is really serious. If all you can find is a line somewhere about ". . .may also be useful for X Disease" or "as an antagonist of Receptor Y", with no more details, then you can ignore that.

The data for specific compounds will vary quite a bit, too. Everyone likes the sorts of patents that list off the compounds and give you real assay numbers next to them. Unfortunately, many filings take the "A, B, C" approach, where they bin the compounds into a few activity classes. That at least lets you pick the more potent ones out from the lesser ones, but there's an even more egregious practice. That's where there's a description of the assay, ending with a line like "All of the compounds claimed showed activity of at least 10 micromolar under these conditions". That sort of stuff drives me crazy, and I really think it should be legally discouraged. I think that this is gradually disappearing from the world, and speed the day.

If you're looking for the best compound in the whole patent, playing "hunt the clinical candidate", well, that can be a fun game. Sometimes it's clear in the way the claim language narrows down to a handful of "most preferred" compounds. Sometimes you can infer it by going through the experimental section and noting when a bunch of 50-milligram procedures suddenly jump to 25-gram procedures (or more). But if a company really wants to hide their single best compound in a forest of other good ones, it can be done. I once had to dig through a pile of Exelixis kinase patents, looking for structures for their clinical candidates, and after about a week, I concluded that it just couldn't be done. And they're not alone.

Comments (24) + TrackBacks (0) | Category: Patents and IP


1. Jose on October 1, 2009 9:16 AM writes...

I have to strongly disagree. Patents are written to be intentionally Byzantine, obstructionist, and misleading. What incentive would any company have to highlight what actually works well? Isn't it better to let the competition flounder among will-o-wisp analogues? They simply cannot be called a reasonable resource. Out of 20 odd preps I've extracted from patents, *one* worked anything like advertised, and it was a fairly simple starting material for a whole series.

Permalink to Comment

2. KristinFromIntellogist on October 1, 2009 9:57 AM writes...

I work in the patent information side of things, so I rarely get the benefit of seeing how someone on the drug development side would read/use a patent document. Great post, Derek!

I wanted to quickly add something for the patent search novices out there. My company offers a free website,, which hosts best practice articles on how to find the patents you are interested in. We also offer reviews and comparisons of available patent search tools, and a discussion forum for anyone with patent search questions.

The site is available entirely free at - hope to see you there!

Permalink to Comment

3. Greg Hlatky on October 1, 2009 10:01 AM writes...

I have to strongly disagree with the above comment. Having read a couple of thousand patents and used examples from a couple of hundred, I've never had any difficulty determining the nature of the invention or the best mode of practice laid out by the applicants or repeating examples provided.

Again, patents aren't journal articles. Applicants are legally obligated to lay out the best mode of practice and describe the field to be claimed. They are not obligated to explain every possible permutation of the invention or deal in hypotheses. No sensible inventor would.

Permalink to Comment

4. er... on October 1, 2009 10:04 AM writes...

"What incentive would any company have to highlight what actually works well? Isn't it better to let the competition flounder among will-o-wisp analogues? They simply cannot be called a reasonable resource."

Jose - I'm not sure your comments hang together... Just because a patent spec may seek to 'hide' the important compounds, doesn't mean the proceedures in the examples won't work. Put another way, if the compound you want to make is exemplified, then the proceedure for making it should be an honest account. If you want to know whether that compound is the lead, that's a different matter.

An important part of any patent is providing an enabling disclosure of the invention (roughly speaking, enough information to enable you to work the invention, e.g. make the compounds etc). There may be a lot of waffle in the general proceedures section, but the examples themselves need to work.

If an academic paper puts forth an unworkable synthesis (which probabaly won't be picked up by peer review unless pretty obviously wrong) the result will, at worst, be embarassment and the need to retract. If the proceedures in a patent don't work you will find that, even if it gets past the patent office, when you try to enforce it, it will simply be declared invalid. If this happens to the patent protecting your current blockbuster, the cost will be slightly worse than a red face (i.e. £££££££!).

I know everyone will quickly jump to the defence of peer review... but 'review' in court involves expert witnesses of the same, if not significantly higher, calibre as any journal reviewer... and some pretty well-trained (and well paid) lawyers will ensure that EVERY detail gets examined to death!

In short, you would have to be clinically insane to knowingly put a non-working prep into a patent you seriously hope to rely on commercially. To put one into a paper, you just need to be pretty lazy and/or arrogant. (I can think of a few academics who fit the bill).

Permalink to Comment

5. Me on October 1, 2009 10:21 AM writes...

I'm a big fan of Crossfire Commander for patent searches. The search results tell you the exact page/example number of the structure you're looking for, and often also list the experimentals and NMR's.

Permalink to Comment

6. Calvin on October 1, 2009 10:34 AM writes...

I have to say that as a graduate student I found patents impossible to understand or navigate through. Now that I'm in industry and have spent a great deal of time reading them I've gotten the hang of it. In fact it was the writing of patents that really sorted me out and now I can whizz through them and find what I want. It definitely takes practice and help from a more experienced chemist to educate you as to wat to look for.
The only thing that really really annoys me is that they tend to be supplied as pdf images, rather than text, so the text becomes unsearchable unless you're willing spend some money on sotfware to (badly) read them. And that definitely seems to be done deliberately to get int he way. Man, that irritates me!

Permalink to Comment

7. labrat on October 1, 2009 10:42 AM writes...

Really it all depends on the company. I have seen excellent patent preps and absolute nightmares. The patent for Tarceva worked as advertised not one problem. However the patent for T-705 is almost written in code. I have never seen a more twisted patent.

Permalink to Comment

8. Anonymous on October 1, 2009 11:10 AM writes...

Always read the claims first.

Permalink to Comment

9. molecular architect on October 1, 2009 11:42 AM writes...

Having played "hunt the clinical candidate" many times, I think this post provides an excellent summary of how to approach patents. It took me several years to get the hang of it.

I see patents as a necessary and useful part of the literature but I hate to read them - even my own!

I have to agree with Calvin (#6). I can't understand why the USPTO's pdf files are images instead of searchable text. After all, the damn things are submitted as word processor files. They have to take extra steps to render them unsearchable.

Permalink to Comment

10. Steve on October 1, 2009 12:31 PM writes...

I've used detailed experimental procedures from patents before, and they've always worked great. The vexing problem is how to get a complete PDF of the entire patent. You can save each page as an individual PDF file, but how can you get a patent as a complete PDF like you can a journal article? Has anyone found a viable workaround for this problem?

Permalink to Comment

11. Greg Hlatky on October 1, 2009 1:13 PM writes...

To download complete patents in PDF, try the European Patent Office website (

Permalink to Comment

12. Cassius on October 1, 2009 1:18 PM writes...


I haven't had much luck with US Patents (they always seem to ask you to purchase them). But for International Patents and applications, getting them through SciFinder is easy. Follow the ChemPort link, and it will open a browser defqaulted to the "Bibliographic Data" tab (you'll see pdf's in the upper right). At the top, click on the "Original Document" tab and it will open the full pdf.

Anyone have suggestions for US patents????

Permalink to Comment

13. Petros on October 1, 2009 1:40 PM writes...

Some useful points there for the uninitiated. But it is exceptionally rare to be able to pick out a canidate compound from the original filing, which may not even include it as a specific example.

Better options are process and crystalline form cases although some companies occassionally let things slip through becuase a global find and replace misses one reference due to a syntax error.

But remember when writing the early specs the aim is to optimize coverage. I always tried to do that when I spoon fed our patent attornies, so it will be difficult to spot the best compound unless a stack of biological data on a single compound are given

Permalink to Comment

14. DerekF on October 1, 2009 2:28 PM writes...

For searchable text of patents, the esp@ce website ( provides not only pdf images but also text files of the specification and claims. These appear to be generated by character recognition from the pdfs, so they're by no means perfect, but they are free and easy to get to. Plus there are translations available (not invariably, and machine translation of limited quality) if the patent is in a common European language. The Japanese Patent Office also provides machine translated text of Japanese patents, but it's pretty much a last resort - look for an equivalent first.

Permalink to Comment

15. DerekF on October 1, 2009 2:31 PM writes...

Oh, and for US patents, the US PTO website ( gets you to the text of US patents and printed publications for many years back. It's searchable; but as we all know, structures beat the heck out of words in trying to find particular compounds.

Permalink to Comment

16. milkshake on October 1, 2009 2:38 PM writes...

We used PatentHunter subscription service to pull patents fulltext but we stopped since FreePatentsOnline came out. This one comes sometimes interspersed with free adds but they is sufficiently inobstructive. Free (non-trial) registration is required if you want to access pdf files with pictures but one only has to do it once

Permalink to Comment

17. Pat Pending on October 1, 2009 3:15 PM writes...

Cassius, for more recent patents Google has complete patents as PDF files. In addition, you can word search the PTO data base and get patents but not as images which are actually tifs not PDFs however you can download an image at a time and then print it out for a complete patent. In the old days you would need to buy at $3.00 a patent the patent from the PTO or have someone copy it from the PTO files or microfilm.

Permalink to Comment

18. Bruce Hamilton on October 1, 2009 3:17 PM writes...

I still use Google Patents for word searching of US patents - it's very fast, and allowing internal searching of patent text online. It also allows easy downloads of PDFs of the patents and applications.

Word searching of chemical patents and applications is important to me, as it helps obtain rapid background information of what others have tried - usually on industrial scales. Some of the advanced search features may help quickly narrow searches. I gave up on the UPPTO site once Google pataents appeared several years ago.

I'm not sure of Google Patents status within Google's Global Domination Plan, as it's no longer appears in some menus, and it's still beta.

Permalink to Comment

19. Anonymous on October 1, 2009 5:32 PM writes...

I've been using Reaxys, which I think is a newer replacement for Crossfire Commander (or, at least, it is a much easier to use web interface to access Beilstein). One of the features I really like about this program is that when you do a search and get a list of reactions (just like in SciFinder), any reaction that came from a patent has a link to the experimental procedure. You click this link and the entire experimental text for that reaction shows up right in the browser. You don't need to go download the patent, search through hundreds of obfuscated pages, and waste all sorts of time. You just click the link and the text is there. I wish it would do this for all references, but, since patents are generally the most annoying, this is great.

I find this to be the most useful feature of Reaxys, so, of course, when we had training for the program, nobody mentioned it. Maybe Crossfire Commander could do this too, but I found that program very non-user-friendly, so I never really learned how to use it.

Permalink to Comment

20. examiner on October 2, 2009 1:33 AM writes...

Steve: Patent Lens is also a good jumping point to get a full PDF of a patent. Some national patent office sites are good. Google patents is alright, but useless if you want an US application. EPOline and Patentscope as well as USPTO have full text html options.

Permalink to Comment

21. anon on October 2, 2009 3:52 AM writes...

Organic Process Research and Development, every now and then, publishes a review of process patents.

How does one look through the last months chemistry patents using either USPTO or ep.espace?

(If that is the appropriate question to ask? If not, what is the appropriate question and answer for what I'm angling at?).

Permalink to Comment

22. Alig on October 2, 2009 7:32 AM writes...

I agree with examiner. Patent Lens is the easiest way to get full PDFs of patents.

Permalink to Comment

23. researchfella on October 3, 2009 11:23 PM writes...

Try (seriously). Set up a user account (it's free), and then you can search for and download entire PDFs of patent documents. If you don't log on with your user account, you can't get the documents, but you can still get limited information.

Permalink to Comment

24. also_anon on October 4, 2009 11:24 AM writes...

For a unix commandline (ie non-windows OS or cygwin) the pat2pdf shell script from works well to create a pdf out of the page images from the USPTO. For a one-off patent there is www.pat2pdf.ORG (NOT COM) which free and based off of the shell script.

They both require you to know the patent number, but that you can get from the USPTO patent search on their website. It's the most convenient patent search since Inventor/Assignee Name/Town are searchable when you want to know what company X has been up to. You can also search by patent classification codes (a rough and archaic keyword system) so you can find references to fans despite their having been referred to as 'Air Movement Devices (AMD)'.

As for patent quality, I have found better preps in the patent lit than the academic lit for some of my starting materials. I found that anything with less than 20 carbons and no sterochem is not sexy enough for academic publications but is often industrially important.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

Gitcher SF5 Groups Right Here
Changing A Broken Science System
One and Done
The Latest Protein-Protein Compounds
Professor Fukuyama's Solvent Peaks
Novartis Gets Out of RNAi
Total Synthesis in Flow
Sweet Reason Lands On Its Face