STORM for live cell imaging

classic Classic list List threaded Threaded
6 messages Options
Aryeh Weiss Aryeh Weiss
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

I am wondering how fast STORM and related single particle localization
methods are.

I would appreciate if list-members who are using these techniques for
live cell imaging would
let us know how fast they image. Is it down to seconds, or is it tens of
seconds to minutes?

Thanks in advance
--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051
0000001ed7f52e4a-dmarc-request 0000001ed7f52e4a-dmarc-request
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****


 
Hello,
 
I have not done live cell localization microscopy, only fixed, but as I could not see other replies, I will try to answer the question. E.g. Jones et al. (2011) Nature Methods 8, 499–505: We achieved a Nyquist resolution of ~20 nm with a time resolution as high as 0.5 s for 2D STORM imaging. Moreover, we achieved 3D volumetric super-resolution imaging of live cells with an overall resolution of 30 nm in the x-y dimension and 50 nm in the z dimension at time resolutions as fast as 1–2 s, albeit with relatively few independent snapshots.

I find this impressive and obviously one has to consider how much motion blurring is acceptable and what resolution should be achieved. But there is one point which is not often mentioned: It is easier to image small dots (0-d, sparse) than for fibers (1-d, sparse) or extended areas (2d, not sparse). The number of necessary localizations to fulfill Nyquist (or go much beyond it as Betzig points out)  is much higher for the more extended objects than for dots or clusters and it takes longer to image these samples. The 'fill-factor' is an important aspect in superesolution imaging.


best wishes

Andreas

 
-----Original Message-----
From: Aryeh Weiss <[hidden email]>
To: CONFOCALMICROSCOPY <[hidden email]>
Sent: Sun, 12 Mar 2017 8:12
Subject: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

I am wondering how fast STORM and related single particle localization
methods are.

I would appreciate if list-members who are using these techniques for
live cell imaging would
let us know how fast they image. Is it down to seconds, or is it tens of
seconds to minutes?

Thanks in advance
--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051
Aryeh Weiss Aryeh Weiss
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Thank you to the members of the list who replied (some on-list, some
off-list). The responses vary from a few seconds to minutes. Of course
it depends on many variables such as whether the objects are extended or
localized, how accurate the localization must be, which dyes are used,
and more.

One common theme is that the fastest acquisition frame rates are 10-30
frames.second (assuming a dye that photo-switches quickly is used).
100-1000 frames are acquired per image, so we have 3-30 seconds as a
lower limit.  In the paper referenced below we have 500 frames/second ,
so 0.5 second/image implies that 250 frames were used to reconstruct
each image (do I have that right?)

--aryeh


On 14/03/2017 0:00, [hidden email] wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
>
>  
> Hello,
>  
> I have not done live cell localization microscopy, only fixed, but as I could not see other replies, I will try to answer the question. E.g. Jones et al. (2011) Nature Methods 8, 499–505: We achieved a Nyquist resolution of ~20 nm with a time resolution as high as 0.5 s for 2D STORM imaging. Moreover, we achieved 3D volumetric super-resolution imaging of live cells with an overall resolution of 30 nm in the x-y dimension and 50 nm in the z dimension at time resolutions as fast as 1–2 s, albeit with relatively few independent snapshots.
>
> I find this impressive and obviously one has to consider how much motion blurring is acceptable and what resolution should be achieved. But there is one point which is not often mentioned: It is easier to image small dots (0-d, sparse) than for fibers (1-d, sparse) or extended areas (2d, not sparse). The number of necessary localizations to fulfill Nyquist (or go much beyond it as Betzig points out)  is much higher for the more extended objects than for dots or clusters and it takes longer to image these samples. The 'fill-factor' is an important aspect in superesolution imaging.
>
>
> best wishes
>
> Andreas
>
>  
> -----Original Message-----
> From: Aryeh Weiss <[hidden email]>
> To: CONFOCALMICROSCOPY <[hidden email]>
> Sent: Sun, 12 Mar 2017 8:12
> Subject: STORM for live cell imaging
>
> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> I am wondering how fast STORM and related single particle localization
> methods are.
>
> I would appreciate if list-members who are using these techniques for
> live cell imaging would
> let us know how fast they image. Is it down to seconds, or is it tens of
> seconds to minutes?
>
> Thanks in advance
> --aryeh
>


--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051
john.oreopoulos john.oreopoulos
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

I think this paper is timely and might be of interest to you regarding  
live-cell super-resolution (depending on your application and the type  
biological process you want to observe, of course). "FLINC" could be  
another interesting tool to add to the super-resolution toolbox:

http://www.nature.com/nmeth/journal/vaop/ncurrent/full/nmeth.4221.html?WT.feed_name=subjects_molecular-biology

John Oreopoulos
Staff Scientist
Andor Technology
www.andor.com


Quoting Aryeh Weiss <[hidden email]>:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Thank you to the members of the list who replied (some on-list, some
> off-list). The responses vary from a few seconds to minutes. Of course
> it depends on many variables such as whether the objects are extended
> or localized, how accurate the localization must be, which dyes are
> used, and more.
>
> One common theme is that the fastest acquisition frame rates are 10-30
> frames.second (assuming a dye that photo-switches quickly is used).
> 100-1000 frames are acquired per image, so we have 3-30 seconds as a
> lower limit.  In the paper referenced below we have 500 frames/second ,
> so 0.5 second/image implies that 250 frames were used to reconstruct
> each image (do I have that right?)
>
> --aryeh
>
>
> On 14/03/2017 0:00, [hidden email] wrote:
>> *****
>> To join, leave or search the confocal microscopy listserv, go to:
>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>> Post images on http://www.imgur.com and include the link in your posting.
>> *****
>>
>>
>>  Hello,
>>  I have not done live cell localization microscopy, only fixed, but  
>>  as I could not see other replies, I will try to answer the  
>> question. E.g. Jones et al. (2011) Nature Methods 8, 499?505: We  
>> achieved a Nyquist resolution of ~20 nm with a time resolution as  
>> high as 0.5 s for 2D STORM imaging. Moreover, we achieved 3D  
>> volumetric super-resolution imaging of live cells with an overall  
>> resolution of 30 nm in the x-y dimension and 50 nm in the z  
>> dimension at time resolutions as fast as 1?2 s, albeit with  
>> relatively few independent snapshots.
>>
>> I find this impressive and obviously one has to consider how much  
>> motion blurring is acceptable and what resolution should be  
>> achieved. But there is one point which is not often mentioned: It  
>> is easier to image small dots (0-d, sparse) than for fibers (1-d,  
>> sparse) or extended areas (2d, not sparse). The number of necessary  
>>  localizations to fulfill Nyquist (or go much beyond it as Betzig  
>> points out)  is much higher for the more extended objects than for  
>> dots or clusters and it takes longer to image these samples. The  
>> 'fill-factor' is an important aspect in superesolution imaging.
>>
>>
>> best wishes
>>
>> Andreas
>>
>>  -----Original Message-----
>> From: Aryeh Weiss <[hidden email]>
>> To: CONFOCALMICROSCOPY <[hidden email]>
>> Sent: Sun, 12 Mar 2017 8:12
>> Subject: STORM for live cell imaging
>>
>> *****
>> To join, leave or search the confocal microscopy listserv, go to:
>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>> Post images on http://www.imgur.com and include the link in your posting.
>> *****
>>
>> I am wondering how fast STORM and related single particle localization
>> methods are.



>>
>> I would appreciate if list-members who are using these techniques for
>> live cell imaging would
>> let us know how fast they image. Is it down to seconds, or is it tens of
>> seconds to minutes?
>>
>> Thanks in advance
>> --aryeh
>>
>
>
> --
> Aryeh Weiss
> Faculty of Engineering
> Bar Ilan University
> Ramat Gan 52900 Israel
>
> Ph:  972-3-5317638
> FAX: 972-3-7384051
George McNamara George McNamara
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi John, thanks for posting the super-resolution FLINC link!

Also published online this week in Nature Methods is an interesting new
optogenetics paper:

http://www.nature.com/nmeth/journal/vaop/ncurrent/full/nmeth.4222.html

doi:10.1038/nmeth.4222

Optogenetic control with a photocleavable protein, PhoCl

Wei Zhang,    Alexander W Lohman,    Yevgeniya Zhuravlova, Xiaocen
Lu,    Matthew D Wiens, Hiofan Hoi,    Sine Yaganoglu, Manuel A Mohr,    
Elena N Kitova,    John S Klassen,    Periklis Pantazis,    Roger J
Thompson    & Robert E Campbell

To expand the range of experiments that are accessible with
optogenetics, we developed a photocleavable protein (PhoCl) that
spontaneously dissociates into two fragments after violet-light-induced
cleavage of a specific bond in the protein backbone. We demonstrated
that PhoCl can be used to engineer light-activatable Cre recombinase,
Gal4 transcription factor, and a viral protease that in turn was used to
activate opening of the large-pore ion channel Pannexin-1.

//

The Pannexin-1 channel may be especially fun, since could let in
ligand(s) that could activate FP biosensors, or conventional FL
reporters like Fura-2, Fluo-4, that are on the inside, a'la Demuro and
Parker optical patch clamping (URl is one of several papers by these
authors)
https://www.ncbi.nlm.nih.gov/pubmed/16103278

enjoy,
George

On 3/14/2017 12:50 PM, [hidden email] wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> I think this paper is timely and might be of interest to you regarding
> live-cell super-resolution (depending on your application and the type
> biological process you want to observe, of course). "FLINC" could be
> another interesting tool to add to the super-resolution toolbox:
>
> http://www.nature.com/nmeth/journal/vaop/ncurrent/full/nmeth.4221.html
>
> John Oreopoulos
> Staff Scientist
> Andor Technology
> www.andor.com
>
>
> Quoting Aryeh Weiss <[hidden email]>:
>
>> *****
>> To join, leave or search the confocal microscopy listserv, go to:
>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>> Post images on http://www.imgur.com and include the link in your
>> posting.
>> *****
>>
>> Thank you to the members of the list who replied (some on-list, some
>> off-list). The responses vary from a few seconds to minutes. Of course
>> it depends on many variables such as whether the objects are extended
>> or localized, how accurate the localization must be, which dyes are
>> used, and more.
>>
>> One common theme is that the fastest acquisition frame rates are 10-30
>> frames.second (assuming a dye that photo-switches quickly is used).
>> 100-1000 frames are acquired per image, so we have 3-30 seconds as a
>> lower limit.  In the paper referenced below we have 500 frames/second ,
>> so 0.5 second/image implies that 250 frames were used to reconstruct
>> each image (do I have that right?)
>>
>> --aryeh
>>
>>
>> On 14/03/2017 0:00, [hidden email] wrote:
>>> *****
>>> To join, leave or search the confocal microscopy listserv, go to:
>>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>> Post images on http://www.imgur.com and include the link in your
>>> posting.
>>> *****
>>>
>>>
>>>  Hello,
>>>  I have not done live cell localization microscopy, only fixed, but
>>>  as I could not see other replies, I will try to answer the  
>>> question. E.g. Jones et al. (2011) Nature Methods 8, 499?505: We  
>>> achieved a Nyquist resolution of ~20 nm with a time resolution as  
>>> high as 0.5 s for 2D STORM imaging. Moreover, we achieved 3D  
>>> volumetric super-resolution imaging of live cells with an overall  
>>> resolution of 30 nm in the x-y dimension and 50 nm in the z  
>>> dimension at time resolutions as fast as 1?2 s, albeit with  
>>> relatively few independent snapshots.
>>>
>>> I find this impressive and obviously one has to consider how much  
>>> motion blurring is acceptable and what resolution should be  
>>> achieved. But there is one point which is not often mentioned: It  
>>> is easier to image small dots (0-d, sparse) than for fibers (1-d,  
>>> sparse) or extended areas (2d, not sparse). The number of necessary
>>>  localizations to fulfill Nyquist (or go much beyond it as Betzig  
>>> points out)  is much higher for the more extended objects than for  
>>> dots or clusters and it takes longer to image these samples. The
>>> 'fill-factor' is an important aspect in superesolution imaging.
>>>
>>>
>>> best wishes
>>>
>>> Andreas
>>>
>>>  -----Original Message-----
>>> From: Aryeh Weiss <[hidden email]>
>>> To: CONFOCALMICROSCOPY <[hidden email]>
>>> Sent: Sun, 12 Mar 2017 8:12
>>> Subject: STORM for live cell imaging
>>>
>>> *****
>>> To join, leave or search the confocal microscopy listserv, go to:
>>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>> Post images on http://www.imgur.com and include the link in your
>>> posting.
>>> *****
>>>
>>> I am wondering how fast STORM and related single particle localization
>>> methods are.
>
>
>
>>>
>>> I would appreciate if list-members who are using these techniques for
>>> live cell imaging would
>>> let us know how fast they image. Is it down to seconds, or is it
>>> tens of
>>> seconds to minutes?
>>>
>>> Thanks in advance
>>> --aryeh
>>>
>>
>>
>> --
>> Aryeh Weiss
>> Faculty of Engineering
>> Bar Ilan University
>> Ramat Gan 52900 Israel
>>
>> Ph:  972-3-5317638
>> FAX: 972-3-7384051

--


George McNamara, PhD
Houston, TX 77054
[hidden email]
https://www.linkedin.com/in/georgemcnamara
https://works.bepress.com/gmcnamara/75/
http://www.ncbi.nlm.nih.gov/myncbi/browse/collection/44962650
Guy Hagen Guy Hagen
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: STORM for live cell imaging

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Dear Colleges:

In traditional PALM/STORM experiments, 10,000-20,000 frames are needed to
achieve a satisfactory reconstruction. If we use an exposure time of 100ms,
this leads to at least 15 minutes to acquire 1 super-resolution image. To
go faster, we need to increase the density of photo-activated molecules in
the sample. This leads to the need for specialized algorithms to process
the data. There has been quite a bit of work in this area, including some
of these examples:

3DenseSTORM:
https://www.osapublishing.org/oe/abstract.cfm?uri=oe-22-25-31263

DAOSTORM:  http://www.nature.com/nmeth/journal/v8/n4/full/nmeth0411-279.html

FALCON:  http://www.nature.com/articles/srep04577

3D FALCON: https://www.osapublishing.org/boe/abstract.cfm?uri=boe-5-11-3935

and several others.

best wishes,
Guy Hagen





On Tue, Mar 14, 2017 at 12:37 PM, George McNamara <[hidden email]
> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hi John, thanks for posting the super-resolution FLINC link!
>
> Also published online this week in Nature Methods is an interesting new
> optogenetics paper:
>
> http://www.nature.com/nmeth/journal/vaop/ncurrent/full/nmeth.4222.html
>
> doi:10.1038/nmeth.4222
>
> Optogenetic control with a photocleavable protein, PhoCl
>
> Wei Zhang,    Alexander W Lohman,    Yevgeniya Zhuravlova, Xiaocen Lu,
> Matthew D Wiens, Hiofan Hoi,    Sine Yaganoglu, Manuel A Mohr,    Elena N
> Kitova,    John S Klassen,    Periklis Pantazis,    Roger J Thompson    &
> Robert E Campbell
>
> To expand the range of experiments that are accessible with optogenetics,
> we developed a photocleavable protein (PhoCl) that spontaneously
> dissociates into two fragments after violet-light-induced cleavage of a
> specific bond in the protein backbone. We demonstrated that PhoCl can be
> used to engineer light-activatable Cre recombinase, Gal4 transcription
> factor, and a viral protease that in turn was used to activate opening of
> the large-pore ion channel Pannexin-1.
>
> //
>
> The Pannexin-1 channel may be especially fun, since could let in ligand(s)
> that could activate FP biosensors, or conventional FL reporters like
> Fura-2, Fluo-4, that are on the inside, a'la Demuro and Parker optical
> patch clamping (URl is one of several papers by these authors)
> https://www.ncbi.nlm.nih.gov/pubmed/16103278
>
> enjoy,
> George
>
> On 3/14/2017 12:50 PM, [hidden email] wrote:
>
>> *****
>> To join, leave or search the confocal microscopy listserv, go to:
>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>> Post images on http://www.imgur.com and include the link in your posting.
>> *****
>>
>> I think this paper is timely and might be of interest to you regarding
>> live-cell super-resolution (depending on your application and the type
>> biological process you want to observe, of course). "FLINC" could be
>> another interesting tool to add to the super-resolution toolbox:
>>
>> http://www.nature.com/nmeth/journal/vaop/ncurrent/full/nmeth.4221.html
>>
>> John Oreopoulos
>> Staff Scientist
>> Andor Technology
>> www.andor.com
>>
>>
>> Quoting Aryeh Weiss <[hidden email]>:
>>
>> *****
>>> To join, leave or search the confocal microscopy listserv, go to:
>>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>> Post images on http://www.imgur.com and include the link in your
>>> posting.
>>> *****
>>>
>>> Thank you to the members of the list who replied (some on-list, some
>>> off-list). The responses vary from a few seconds to minutes. Of course
>>> it depends on many variables such as whether the objects are extended
>>> or localized, how accurate the localization must be, which dyes are
>>> used, and more.
>>>
>>> One common theme is that the fastest acquisition frame rates are 10-30
>>> frames.second (assuming a dye that photo-switches quickly is used).
>>> 100-1000 frames are acquired per image, so we have 3-30 seconds as a
>>> lower limit.  In the paper referenced below we have 500 frames/second ,
>>> so 0.5 second/image implies that 250 frames were used to reconstruct
>>> each image (do I have that right?)
>>>
>>> --aryeh
>>>
>>>
>>> On 14/03/2017 0:00, [hidden email] wrote:
>>>
>>>> *****
>>>> To join, leave or search the confocal microscopy listserv, go to:
>>>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>>> Post images on http://www.imgur.com and include the link in your
>>>> posting.
>>>> *****
>>>>
>>>>
>>>>  Hello,
>>>>  I have not done live cell localization microscopy, only fixed, but  as
>>>> I could not see other replies, I will try to answer the  question. E.g.
>>>> Jones et al. (2011) Nature Methods 8, 499?505: We  achieved a Nyquist
>>>> resolution of ~20 nm with a time resolution as  high as 0.5 s for 2D STORM
>>>> imaging. Moreover, we achieved 3D  volumetric super-resolution imaging of
>>>> live cells with an overall  resolution of 30 nm in the x-y dimension and 50
>>>> nm in the z  dimension at time resolutions as fast as 1?2 s, albeit with
>>>> relatively few independent snapshots.
>>>>
>>>> I find this impressive and obviously one has to consider how much
>>>> motion blurring is acceptable and what resolution should be  achieved. But
>>>> there is one point which is not often mentioned: It  is easier to image
>>>> small dots (0-d, sparse) than for fibers (1-d,  sparse) or extended areas
>>>> (2d, not sparse). The number of necessary  localizations to fulfill Nyquist
>>>> (or go much beyond it as Betzig  points out)  is much higher for the more
>>>> extended objects than for  dots or clusters and it takes longer to image
>>>> these samples. The 'fill-factor' is an important aspect in superesolution
>>>> imaging.
>>>>
>>>>
>>>> best wishes
>>>>
>>>> Andreas
>>>>
>>>>  -----Original Message-----
>>>> From: Aryeh Weiss <[hidden email]>
>>>> To: CONFOCALMICROSCOPY <[hidden email]>
>>>> Sent: Sun, 12 Mar 2017 8:12
>>>> Subject: STORM for live cell imaging
>>>>
>>>> *****
>>>> To join, leave or search the confocal microscopy listserv, go to:
>>>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>>> Post images on http://www.imgur.com and include the link in your
>>>> posting.
>>>> *****
>>>>
>>>> I am wondering how fast STORM and related single particle localization
>>>> methods are.
>>>>
>>>
>>
>>
>>
>>>> I would appreciate if list-members who are using these techniques for
>>>> live cell imaging would
>>>> let us know how fast they image. Is it down to seconds, or is it tens of
>>>> seconds to minutes?
>>>>
>>>> Thanks in advance
>>>> --aryeh
>>>>
>>>>
>>>
>>> --
>>> Aryeh Weiss
>>> Faculty of Engineering
>>> Bar Ilan University
>>> Ramat Gan 52900 Israel
>>>
>>> Ph:  972-3-5317638
>>> FAX: 972-3-7384051
>>>
>>
> --
>
>
> George McNamara, PhD
> Houston, TX 77054
> [hidden email]
> https://www.linkedin.com/in/georgemcnamara
> https://works.bepress.com/gmcnamara/75/
> http://www.ncbi.nlm.nih.gov/myncbi/browse/collection/44962650
>
Loading...