[ome-devel] FW: Opera Flex file lattest changes
Melissa Linkert
melissa at glencoesoftware.com
Thu Apr 3 01:37:14 BST 2014
Hi Ghislain,
> We now have a proposed fix for this issue:
>
> https://github.com/openmicroscopy/bioformats/pull/1018
>
> That's unfortunately unlikely to make it into the imminent 5.0.1 release, but I would expect it to be in 5.0.2.
> [G.B.] Is the loci_tools.jar from the lattest build incorporating this?
No, not yet. Once that is marked as being "closed"/"merged", it will be
included in the files in the "Latest build" column of:
http://downloads.openmicroscopy.org/bio-formats/5.0.0
> > Finally, if you can tell me what is the best way to tell the reader to open the images as a well Image (ie. group images by well), that would be useful. In general I would like to get the properties for the entire well (ie. the series count should match the number of field in the well).
>
> There unfortunately isn't really a way to do that - you can either get the entire plate (the default) or a single field in a well (by calling
> setGroupFiles(false) on the reader).
> [G.B.] Do you think it would be possible to get a grouping mechanism in place which allows to specify the aggregation. This would be particularly useful for HCI type data
That's not really our intention at the moment, as the default behaviour of
setId gives enough information to selectively read a subset of the wells/fields.
If there are measurable performance problems with the approach outlined
below, then we can certainly consider alternatives.
> You can, however, extract the various plate/well dimensions from the reader's metadata object and use that to pull out the relevant images:
> [G.B.] From the code below, it would seem that I can access all of the dimensions from the well using a single image. Perhaps I can use this instead.
OK, great - if you run into any problems with that code, please let us know.
Regards,
-Melissa
On Tue, Apr 01, 2014 at 01:04:25AM +0000, Bonamy, Ghislain wrote:
> Hi Mellisa,
>
> Thanks for the prompt response! A few quick questions, below:
>
> Ghislain Bonamy, PhD
>
>
> -----Original Message-----
> From: Melissa Linkert [mailto:melissa.linkert at gmail.com] On Behalf Of Melissa Linkert
> Sent: Tuesday, April 01, 2014 5:22 AM
> To: Bonamy, Ghislain
> Cc: ome-devel at lists.openmicroscopy.org.uk
> Subject: Re: FW: Opera Flex file lattest changes
>
> Hi Ghislain,
>
> > Just wanted to let you know I upgraded my package to the bioformats 5 release (loci_tools.jar).
> >
> > The problem persists. I therefore assume that when PE changed the file format something broke the reader function and now sizeC (and possibly other metada are broken, like sizeT, sizeZ). The new format for the Opera, PE is now splitting each well into multiple images each image corresponds to a different field which then contains all of the other dimensions (ie. C, Z, T). The field ID, is specified by the last 3 digits of the filename.
>
> We now have a proposed fix for this issue:
>
> https://github.com/openmicroscopy/bioformats/pull/1018
>
> That's unfortunately unlikely to make it into the imminent 5.0.1 release, but I would expect it to be in 5.0.2.
> [G.B.] Is the loci_tools.jar from the lattest build incorporating this?
> [G.B.]
> > Finally, if you can tell me what is the best way to tell the reader to open the images as a well Image (ie. group images by well), that would be useful. In general I would like to get the properties for the entire well (ie. the series count should match the number of field in the well).
>
> There unfortunately isn't really a way to do that - you can either get the entire plate (the default) or a single field in a well (by calling
> setGroupFiles(false) on the reader).
> [G.B.] Do you think it would be possible to get a grouping mechanism in place which allows to specify the aggregation. This would be particularly useful for HCI type data
>
> You can, however, extract the various plate/well dimensions from the reader's metadata object and use that to pull out the relevant images:
> [G.B.] From the code below, it would seem that I can access all of the dimensions from the well using a single image. Perhaps I can use this instead.
>
> // -- begin --
>
> // initialize the file
> ImageReader reader = new ImageReader();
> IMetadata meta = MetadataTools.createOMEXMLMetadata();
> reader.setMetadataStore(meta);
> reader.setId("/path/to/001001001.flex");
>
> // get the dimensions of the plate
> int plateIndex = 0;
> int wellCount = meta.getWellCount(plateIndex); for (int well=0; well<wellCount; well++) {
> int wellRow = meta.getWellRow(plateIndex, well).getValue();
> int wellColumn = meta.getWellColumn(plateIndex, well).getValue();
>
> int fieldCount = meta.getWellSampleCount(plateIndex, well);
> for (int field=0; field<fieldCount; field++) {
> int fieldID = meta.getWellSampleImageRef(plateIndex, well, field);
>
> // find the corresponding Image index (i.e. series) and read the planes
> for (int image=0; image<meta.getImageCount(); image++) {
> if (meta.getImageID(image).equals(fieldID)) {
> reader.setSeries(image);
> for (int plane=0; plane<reader.getImageCount(); plane++) {
> reader.openBytes(plane);
> }
> break;
> }
> }
> }
>
> }
>
> reader.close();
> // -- end --
>
> Regards,
> -Melissa
>
> On Fri, Mar 28, 2014 at 03:46:01AM +0000, Bonamy, Ghislain wrote:
> > Melissa,
> >
> > Just wanted to let you know I upgraded my package to the bioformats 5 release (loci_tools.jar).
> >
> > The problem persists. I therefore assume that when PE changed the file format something broke the reader function and now sizeC (and possibly other metada are broken, like sizeT, sizeZ). The new format for the Opera, PE is now splitting each well into multiple images each image corresponds to a different field which then contains all of the other dimensions (ie. C, Z, T). The field ID, is specified by the last 3 digits of the filename.
> >
> > Finally, if you can tell me what is the best way to tell the reader to open the images as a well Image (ie. group images by well), that would be useful. In general I would like to get the properties for the entire well (ie. the series count should match the number of field in the well).
> >
> > Best,
> >
> > Ghislain Bonamy, PhD
> >
> >
> > -----Original Message-----
> > From: Melissa Linkert [mailto:melissa.linkert at gmail.com] On Behalf Of
> > Melissa Linkert
> > Sent: Friday, March 28, 2014 9:58 AM
> > To: Bonamy, Ghislain
> > Subject: Re: FW: Opera Flex file lattest changes
> >
> > Hi Ghislain,
> >
> > > I am unsure if the emails copied below reached you or not. I hope that you will be able to help me out.
> >
> > My apologies - your messages did reach me, I've just had limited availability over the last couple of weeks.
> >
> > We are investigating this now, and I will follow up on the mailing list once we have more information.
> >
> > Regards,
> > -Melissa
> >
> > On Mon, Mar 24, 2014 at 05:47:22AM +0000, Bonamy, Ghislain wrote:
> > > Hi Melissa,
> > >
> > > I am unsure if the emails copied below reached you or not. I hope that you will be able to help me out.
> > >
> > > Best,
> > >
> > > Ghislain Bonamy, PhD
> > >
> > > From: Bonamy, Ghislain
> > > Sent: Thursday, March 20, 2014 7:30 PM
> > > To: 'melissa.linkert at gmail.com'
> > > Subject: RE: Opera Flex file lattest changes
> > >
> > > Hi Melissa,
> > >
> > > I wanted to also mention that since the new format of PE was
> > > released It would appear that the reader does not correctly handle "reader.getSizeC()"
> > >
> > > I assume that the other dimensions are also incorrect.
> > >
> > > The data is quite big, so I am sending you the images from 2 fields (there are a total of 12 fields) and each fields encodes for 2 channels. There is no Z stacks or T-course... Hope this helps and you can help me get the reader working again.
> > >
> > >
> > > Ghislain Bonamy, PhD
> > >
> > > From: Bonamy, Ghislain
> > > Sent: Thursday, March 20, 2014 10:51 AM
> > > To: 'melissa.linkert at gmail.com'
> > > Subject: RE: Opera Flex file lattest changes
> > >
> > > Hi Melissa,
> > >
> > > Hope this emails find you well. It has been a while since our last discussion. I was hoping you could help me with the request below! Not sure if this is your email address. I seem to have lost your contact details when I moved to my new position.
> > >
> > > Best,
> > >
> > > Ghislain Bonamy, PhD
> > >
> > > From: Bonamy, Ghislain
> > > Sent: Monday, March 17, 2014 7:28 PM
> > > To: 'ome-devel at lists.openmicroscopy.org.uk'
> > > Subject: Opera Flex file lattest changes
> > >
> > > Hi everyone,
> > >
> > > It has been a while since I last posted. I had a quick question, which Melissa can probably answer. The flex file form changed slightly in v2.0 of Evoshell. In particular the file structure. The microscope now generates the files as 1 file per field (as opposed to one file per image). As before the rest of the dimensions (ie. Z-stacks and channels) are encoded into the same file.
> > >
> > > I was wondering how I could request the metadata for a full well in s ingle call? In particular I would like to make our webservice that retrieves the images compatible with the previous version with as little changes as possible. If there was a way to do things like:
> > >
> > > 1) Get the number of fields for Well B01 (002001*)
> > >
> > > 2) Get the image slice for WellB01 Channel2, Z-stack
> > >
> > > That would be perfect.
> > >
> > > Thanks in advance for the help. Hopefully this is clear, otherwise I can post more details.
> > >
> > > Best,
> > >
> > > Ghislain Bonamy, PhD
> > > __________________________________________
> > > Group Leader
> > >
> > > Novartis
> > > Institute for
> > > Tropical
> > > Diseases
> >
> >
More information about the ome-devel
mailing list