Skip to content

Latest commit

 

History

History
89 lines (75 loc) · 4.71 KB

protocol.md

File metadata and controls

89 lines (75 loc) · 4.71 KB
Raw

Step 1: Generate a "hole map"

  • Use about 1 million holes over a 512x512 CCD (for a coverage of approximately 4 holes/pixel; with my grid of holes, this was about 5,000 frames, but your mileage may vary). As best as I can tell, it's better to move the grid of holes randomly, and by relatively large amounts (i.e. 2-3 inter-hole distances), rather than doing a raster-scan of the CCD. There's some speculation that this is related to charge buildup on the CCD, but there's absolutely no evidence that's actually the cause.

  • As an interesting side note, the NTFS file system (and possibly others) bogs down when you have many, many items in the same folder. I tend to keep a folder of folders, each of which contains 1,000 images.

  • Use the python script Processing.py to localize spots in each image. The default options typically work if you start in the folder containing the folder-of-folders. Otherwise, you may need to use the --specification flag to tell the program where to look.

  • Use PostProcessing.py to turn the matlab file generated by Processing (will typically be the day's date in MMDD-map.mat format) into a first-pass "hole map".

Step 2: Imaging

  • Capture your two-color movies.

  • Using fluorescent beads that are roughly equally bright in both channels, capture movies of approximately 300-500 beads, leaving each bead stable for ~10-15 frames. Depending on the stability of your system, you may want to do this fairly frequently (I typically do two rounds of protein imaging, then a single round of bead imaging, then two more rounds of protein.

Step 3: Data-processing

  • Localize beads using WHTrackHighRes. Select only beads that are bright, but not saturated, and well separated from each other.

  • This process is very tedious, so I came up with a couple scripts which, combined with the CellCounter plugin in ImageJ, makes it only slightly less tedious.

    • For one frame in a series, click on every object of interest in the upper half-field using a Type 2 marker in the Cell Counter. In the current version of the software, the lower half can also be selected using Type 1 markers, but if it isn't , will be automatically calculated using the input map. This is a huge time saver.

    • For the rest of the frames in that series, click at least once with a type 2 marker, anywhere.

    • Make sure to leave at least 2-3 completely unmarked frames between each series.

    • Use the CellCounter to save the XML file. I almost always accept the default name.

    • Turn the XML file that CellCounter saves into a WHTrackHighRes-compatible spotlist using XML2Spotlist. From IPython, it will look something like this: run XML2spotlist -m 0916-map_1_20 Red-s1_1.xml

    • Actually track the spots. Through some tinkering, I've put together ProcessAll.m, a matlab script that will call WHTrackHighRes on all the spotlists in the current folder with appropriate parameters. This can take a while to run, but requires no human input. If you aren't using ProcessAll, just run WHTrackDefaults to generate a defaults.mat, and then run WHTrackHighRes on each spotlist. The settings for WHTrackDefaults are:

      • Method: Gauss2DJILA I'm not certain there's a difference between the different tracking methods, but Processing.py has been tested and gives the same output as this method.
      • PixelSize This will depend on your microscope. On the Yildiz Lab Scope in Stanley B347, it's 106.667 nm
      • PixelPadding For tracking beads, I tend to make this a little bit larger, to ensure that it will include the spot in the pixel (and I also make sure not to select beads that are too close together. For protein, you have less control about how close adjacent spots are, so a lower setting (like 7) seems to work fine.

  • For each bead image, generate an "offset map". Use CombineWHSpots on the output for each Bead movie, and when asked if you want to save the output, type y. This should generate another map file, and if the name of the beads folder was Red-s1_1, would then be called offsetsRed-s1_1_N (where N is a number indicating the order of the offset fit, and will be higher with more data).

  • Now, using the hole map which should be good for at least the day, and the offset maps, you should be able to colocalize your objects of interest using CombineWHSpots. The calling syntax will be something like run CombineWHSpots -F -m 0703-map_1_20 -2 offsetsRed-s1_1_7 Dyn99-s1_1/ The -F flag will prevent averaging each point over multiple frames.