michaeljohndolan / fish_analysis Goto Github PK

{\rtf1\ansi\ansicpg1252\cocoartf1561\cocoasubrtf600
{\fonttbl\f0\fswiss\fcharset0 ArialMT;}
{\colortbl;\red255\green255\blue255;\red0\green0\blue0;}
{\*\expandedcolortbl;;\cssrgb\c0\c0\c0;}
\margl1440\margr1440\vieww13080\viewh14820\viewkind0
\deftab720
\pard\pardeftab720\sl280\partightenfactor0

\f0\b\fs24 \cf2 \expnd0\expndtw0\kerning0
\ul \ulc2 Protocol for unprocessed identification of Fcrls cells: \
\
Note: \ulnone Ensure image names match the format: DateOfExp_brain#_slice#_tile#_fcrls_ccl3.nd2
\b0 \

\fs36 0. Directory structure: \
	a. Main directory with macros and code \
	b. Subdirectories for each channel (eg. Ccl3/ and Fcrls/), each subdirectory has a TestData and a Segmentation directory \
	c. RawData directory \
	d. Output directory\
\
1. In the raw data, using Fiji\'92s square selection and clear/clear outside commands, crop out regions outside the brain, hairs, over saturated pixels and other aberrations in the original file. \
2. Save this as a tif, Prefixed with cropped (cropped_201903etc) in the raw data folder. \
3. Split the channels in cropped files and save in the appropriate directories (eg. Ccl3/ and Fcrls/) in the TestData subdirectory  \
4. If not already done, create a ilp project and train a classifier on the TestData \
5.  Open eg. Fcrls_hemi.ilp and use the trained algorithm to identify cells in TestData. Run Ilastik batch mode to create simple segmentations and tifs. Save these in Segmentation directory. \
6. Run Ilastik on all channel directories. \
5. Run the \'93simple_segmentation_postprocess.ijm\'94 ImageJ macro on the segmentations of each channel. This will save postprocessed and DISPLAY images (with artificially dilated cells for easy viewing) in the Output directory. \
6. Open R and run overlap_analysis.R which will call the the overlap_analysis.ijm macro and extract colocalized cells and numbers. May require some tweaking of code depending on the images. \
\
\
\
}