From owner-chemistry@ccl.net Mon Dec 13 11:04:00 2010 From: "Peter Schmidtke pschmidtke^^^mmb.pcb.ub.es" To: CCL Subject: CCL: Prediction & Identification of protein-ligand binding sites Message-Id: <-43369-101213104649-22802-i/kfCa5cD/vluDe5p8mvbw|-|server.ccl.net> X-Original-From: Peter Schmidtke Content-Type: multipart/alternative; boundary=Apple-Mail-1-415370384 Date: Mon, 13 Dec 2010 16:46:22 +0100 Mime-Version: 1.0 (Apple Message framework v1081) Sent to CCL by: Peter Schmidtke [pschmidtke ~ mmb.pcb.ub.es] --Apple-Mail-1-415370384 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=iso-8859-1 Very good point. However, one has to bear in mind that despite the fact = that docking software can produce reasonable poses, the scoring schemes = are still not reliable enough. So as you said, a consensus prediction = between docking and pocket prediction can give hints. But there again = it's not proof.=20 Ideally, from the output of the previous two analyses you could run MD = simulations then with your ligand docked into the pocket to see if it is = stable...which is obviously a lot heavier, but certainly less than = trying to get a crystal structure ;) On 13/12/2010, at 11:34, miriam sgobba miriam.sgobba||gmail.com wrote: > Dear Deepangi, >=20 > if you know from experimental data one or more compounds that bind = your protein, you can also try the blind docking technique, which allow = you to perform docking simulations on large surfaces of the protein in = order to get unbiased mapping of the ligand.=20 > For references, see: > = http://www.sciencedirect.com/science?_ob=3DArticleURL&_udi=3DB6T36-4J5D6GW= -5&_user=3D126523&_coverDate=3D02%2F20%2F2006&_rdoc=3D1&_fmt=3Dhigh&_orig=3D= search&_origin=3Dsearch&_sort=3Dd&_docanchor=3D&view=3Dc&_acct=3DC00001035= 8&_version=3D1&_urlVersion=3D0&_userid=3D126523&md5=3D9fb1be8e42cf768ac87b= 921ab39d056b&searchtype=3Da >=20 > http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2373668/?tool=3Dpubmed >=20 > and, for some blind docking applications: >=20 > http://www.springerlink.com/content/p237tl40785m1tu5/ > http://pubs.acs.org/doi/abs/10.1021/ci1001857 >=20 > Usually, a blind docking simulation is followed by a focused docking = and post-docking refinement if you want to better characterize > the ligand binding mode. > In my experience, a consensus among binding site identification = methods (both geometrical and energetical) + blind docking technique is = the best choice to identify a ligand binding site.=20 >=20 > Hope this helps! > Good luck! >=20 > Miriam >=20 >=20 > 2010/12/12 Deepangi Pandit deepangi.pandit++gmail.com = >=20 >=20 > Sent to CCL by: Deepangi Pandit [deepangi.pandit|,|gmail.com] > Thank you Peter, for your detailed reply. I would just like to clarify > my case based on your questions. >=20 > > 1 : Do you want to apply this method on a database of proteins or = just a few structures? >=20 > I am looking at just a few structures to identify the putative drug > binding site. I have PDB files of protein structures (of course, > without the compound) and have experimental data which suggests the > compound is binding but do not know where. >=20 > > 2 : Do you want to identify putative drug binding sites or whatever = binding site a method could find (general purpose pocket prediction) >=20 > I am trying to identify the putative drug binding sites not general > purpose pocket prediction >=20 >=20 > > 3 : Do you want to find conserved pockets among homologs? >=20 > This is not my primary goal. >=20 > Best, > Deepa >=20 > On Sat, Dec 11, 2010 at 4:45 PM, Peter Schmidtke > pschmidtke[-]mmb.pcb.ub.es wrote: > > > > Sent to CCL by: Peter Schmidtke [pschmidtke[*]mmb.pcb.ub.es] > > Dear Deepangi, > > > > protein-ligand binding site detection is a very active field and = there > > is a large variety of methods out there. > > Since the publication of PocketPicker > > (http://www.journal.chemistrycentral.com/content/1/1/7/abstract/) = lots > > of newer methods used the comparative table that was published with > > PocketPicker (table 2). A very recent example is the paper by Andrea > > Volkamer, where you have an update of this table with a few more = recent > > methods (http://pubs.acs.org/doi/full/10.1021/ci100241y). > > > > This comparison is definitely arguable, but at least it gives some = hint > > on general pocket prediction performance. In order to find the = perfect > > method for your purpose you should maybe clarify what is the > > functionality your are seeking. There are lots of webservers out = there, > > methods you have to pay and also free methods. > > > > Some questions that could influence your choice : > > > > 1 : Do you want to apply this method on a database of proteins or = just a > > few structures? > > 2 : Do you want to identify putative drug binding sites or whatever > > binding site a method could find (general purpose pocket prediction) > > 3 : Do you want to find conserved pockets among homologs? > > > > Concerning question 1, if you just need a method for a few = structures > > you can go for slower programs (mostly energy based algorithms) like > > SiteMap from Schr=F6dinger (but you need a license) Q-SiteFinder or > > PocketPicker or Vice (I think it's sold by Tripos) There are surely = more > > than that. > > If you need to process a large number of structures, considering > > geometry based methods might become an interesting choice. Choices = could > > be Ligsite, fpocket, PASS, SiteFinder (included in MOE by CCG) ... > > > > Regarding question 2. Pocket prediction is not equal drug binding = site > > prediction. Recent druggability prediction methods have been = published > > and some scores are available ready for use in SiteMap (I don't know = if > > you still have to calculate it or if they finally included it into = the > > SiteMap distribution) and fpocket. I suspect also QSiteFinder might = give > > indications on druggability (I extrapolate from the methodology I = just > > read very quickly, so maybe I'm wrong here ;) ). > > > > Regarding question 3, I think ligsite csc can help there > > (http://projects.biotec.tu-dresden.de/pocket/), or else hpocket, = which > > is part of the fpocket webserver : > > http://bioserv.rpbs.univ-paris-diderot.fr/fpocket/ > > > > I hope that this gives you at least a direction, there are really = lots > > of methods and it is indeed not easy to know which one to use ;) > > > > Ah, I nearly forgot, metapocket2.0 is a sort of consensus pocket > > prediction method that makes use of LIGSITEcsc, PASS, Q-SiteFinder, > > SURFNET, Fpocket, GHECOM and ConCavity. This has again advantages = and > > disadvantages, but well, at least you know that it exists ;) > > > > Good luck. > > > > Peter Schmidtke > > > > > > > > > > On 12/11/2010 06:44 PM, Deepangi Pandit deepangi.pandit~!~gmail.com = wrote: > >> Sent to CCL by: Deepangi Pandit [deepangi.pandit(!)gmail.com] > >> Dear All: > >> I tried to search literature and Internet to find a reliable method > >> that I can use to identify protein-ligand binding site. I came = across > >> number of papers but I am unable to evaluate the methods before I > >> start using them. I was also unable to find a review which compares > >> the methods. It seems method can be broadly classified as "Geometry > >> based" and "Energy Based". Could you please recommend from your own > >> experience which method gives reasonable results. Also, feel free = to > >> share pros and cons of the method. > >> > >> Thank you. > >> Deepa > >> > >> List of methods I came across > >> > >> Q-Site Finder > >> http://www.bioinformatics.leeds.ac.uk/qsitefinder > >> > >> Surface triplet propensities > >> http://opus.bch.ed.ac.uk/stp > >> > >> Screen > >> http://interface.bioc.columbia.edu/screen> => => => => > > > > >=20 >=20 >=20 > -=3D This is automatically added to each message by the mailing script = =3D- >=20 >=20 >=20 > E-mail to subscribers: CHEMISTRY#,#ccl.net or use:>=20 > E-mail to administrators: CHEMISTRY-REQUEST#,#ccl.net or use>=20>=20>=20> Conferences: = http://server.ccl.net/chemistry/announcements/conferences/ >=20>=20 >=20>=20>=20 >=20 >=20 >=20 Peter Schmidtke ----------------- PhD Student Department of Physical Chemistry School of Pharmacy University of Barcelona Barcelona, Spain --Apple-Mail-1-415370384 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=iso-8859-1 Very = good point. However, one has to bear in mind that despite the fact that = docking software can produce reasonable poses, the scoring schemes are = still not reliable enough. So as you said, a consensus prediction = between docking and pocket prediction can give hints. But there again = it's not proof. 

Ideally, from the output of the = previous two analyses you could run MD simulations then with your ligand = docked into the pocket to see if it is stable...which is obviously a lot = heavier, but certainly less than trying to get a crystal structure = ;)


On 13/12/2010, at 11:34, = miriam sgobba miriam.sgobba||gmail.com wrote:

Dear = Deepangi,

if you know from experimental data one or more=20 compounds that bind your protein, you can also try the blind docking=20 technique, which allow you to perform docking simulations on large=20 surfaces of the protein in order to get unbiased mapping of the ligand. =
For references, see:
http://www.sciencedirect.com/science?_ob=3DArticleURL&am= p;_udi=3DB6T36-4J5D6GW-5&_user=3D126523&_coverDate=3D02%2F20%2F200= 6&_rdoc=3D1&_fmt=3Dhigh&_orig=3Dsearch&_origin=3Dsearch&am= p;_sort=3Dd&_docanchor=3D&view=3Dc&_acct=3DC000010358&_ver= sion=3D1&_urlVersion=3D0&_userid=3D126523&md5=3D9fb1be8e42cf76= 8ac87b921ab39d056b&searchtype=3Da

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2373668/?too= l=3Dpubmed

and, for some blind docking applications:

http://www.springerlink.com/content/p237tl40785m1tu5/
http://pubs.acs.org/doi/abs/10.1021/ci1001857
Usually, a blind docking simulation is followed by a focused docking and=20 post-docking refinement if you want to better characterize
the ligand binding mode.
In my experience, a consensus among binding=20= site identification methods (both geometrical and energetical) + blind=20= docking technique is the best choice to identify a ligand binding=20 site.

Hope this helps!
Good luck!

Miriam


2010/12/12 Deepangi Pandit = deepangi.pandit++gmail.com <owner-chemistry#,#ccl.net>


Sent to CCL by: Deepangi Pandit [deepangi.pandit|,|gmail.com]
Thank you Peter, for your detailed reply. I would just like to = clarify
my case based on your questions.

> 1 : Do you want to apply this method on a database of proteins or = just a few structures?

I am looking at just a few structures to identify the putative drug
binding site. I have PDB files of protein structures (of course,
without the compound) and have experimental data which suggests the
compound is binding but do not know where.

> 2 : Do you want to identify putative drug binding sites or whatever = binding site a method could find (general purpose pocket prediction)

I am trying to identify the putative drug binding sites not general
purpose pocket prediction


> 3 : Do you want to find conserved pockets among homologs?

This is not my primary goal.

Best,
Deepa

On Sat, Dec 11, 2010 at 4:45 PM, Peter Schmidtke
pschmidtke[-]mmb.pcb.ub.es <owner-chemistry|-|ccl.net> wrote:
>
> Sent to CCL by: Peter Schmidtke [pschmidtke[*]mmb.pcb.ub.es]
> Dear Deepangi,
>
> protein-ligand binding site detection is a very active field and = there
> is a large variety of methods out there.
> Since the publication of PocketPicker
> (http://www.journal.chemistrycentral.com/content/1/1/7/ab= stract/) lots
> of newer methods used the comparative table that was published = with
> PocketPicker (table 2). A very recent example is the paper by = Andrea
> Volkamer, where you have an update of this table with a few more = recent
> methods (http://pubs.acs.org/doi/full/10.1021/ci100241y).
= >
> This comparison is definitely arguable, but at least it gives some = hint
> on general pocket prediction performance. In order to find the = perfect
> method for your purpose you should maybe clarify what is the
> functionality your are seeking. There are lots of webservers out = there,
> methods you have to pay and also free methods.
>
> Some questions that could influence your choice :
>
> 1 : Do you want to apply this method on a database of proteins or = just a
> few structures?
> 2 : Do you want to identify putative drug binding sites or = whatever
> binding site a method could find (general purpose pocket = prediction)
> 3 : Do you want to find conserved pockets among homologs?
>
> Concerning question 1, if you just need a method for a few = structures
> you can go for slower programs (mostly energy based algorithms) = like
> SiteMap from Schr=F6dinger (but you need a license) Q-SiteFinder = or
> PocketPicker or Vice (I think it's sold by Tripos) There are surely = more
> than that.
> If you need to process a large number of structures, = considering
> geometry based methods might become an interesting choice. Choices = could
> be Ligsite, fpocket, PASS, SiteFinder (included in MOE by CCG) = ...
>
> Regarding question 2. Pocket prediction is not equal drug binding = site
> prediction. Recent druggability prediction methods have been = published
> and some scores are available ready for use in SiteMap (I don't = know if
> you still have to calculate it or if they finally included it into = the
> SiteMap distribution) and fpocket. I suspect also QSiteFinder might = give
> indications on druggability (I extrapolate from the methodology I = just
> read very quickly, so maybe I'm wrong here ;) ).
>
> Regarding question 3, I think ligsite csc can help there
> (http://projects.biotec.tu-dresden.de/pocket/), or = else hpocket, which
> is part of the fpocket webserver :
> http://bioserv.rpbs.univ-paris-diderot.fr/fpocket/ >
> I hope that this gives you at least a direction, there are really = lots
> of methods and it is indeed not easy to know which one to use = ;)
>
> Ah, I nearly forgot, metapocket2.0 is a sort of consensus = pocket
> prediction method that makes use of LIGSITEcsc, PASS, = Q-SiteFinder,
> SURFNET, Fpocket, GHECOM and ConCavity. This has again advantages = and
> disadvantages, but well, at least you know that it exists ;)
>
> Good luck.
>
> Peter Schmidtke
>
>
>
>
> On 12/11/2010 06:44 PM, Deepangi Pandit deepangi.pandit~!~gmail.com wrote:
>> Sent to CCL by: Deepangi Pandit [deepangi.pandit(!)gmail.com]
>> Dear All:
>> I tried to search literature and Internet to find a reliable = method
>> that I can use to identify protein-ligand binding site. I came = across
>> number of papers but I am unable to evaluate the methods before = I
>> start using them. I was also unable to find a review which = compares
>> the methods. It seems method can be broadly classified as = "Geometry
>> based" and "Energy Based". Could you please recommend from your = own
>> experience which method gives reasonable results. Also, feel = free to
>> share pros and cons of the method.
>>
>> Thank you.
>> Deepa
>>
>> List of methods I came across
>>
>> Q-Site Finder
>> http://www.bioinformatics.leeds.ac.uk/qsitefinder >>
>> Surface triplet propensities
>> http://opus.bch.ed.ac.uk/stp
>>
>> Screen
>> http://interface.bioc.columbia.edu/screen> =      http://www.ccl.net/cgi-bin/ccl/send_ccl_message> =      http://www.ccl.net/cgi-bin/ccl/send_ccl_message> =      http://www.ccl.net/chemistry/sub_unsub.shtml> =      http://www.ccl.net/spammers.txt>
>
>



-=3D This is automatically added to each message by the mailing script = =3D-



E-mail to subscribers: CHEMISTRY#,#ccl.net or use:
     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST#,#ccl.net or use
     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

Subscribe/Unsubscribe:
     http://www.ccl.net/chemistry/sub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemistry/announcements/conference= s/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml

     http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/<= br>




Peter = Schmidtke

-----------------
PhD = Student
Department of Physical Chemistry
School of = Pharmacy
University of Barcelona
Barcelona, = Spain

= --Apple-Mail-1-415370384--