Mike Mikailov dc353f60cd Add files via upload | 4 years ago | |
---|---|---|
docs | 4 years ago | |
heatmap_stitch | 4 years ago | |
image_patch_extract | 4 years ago | |
prediction | 4 years ago | |
LICENSE.md | 4 years ago | |
config_generic.txt | 4 years ago | |
config_generic_cn_true.txt | 4 years ago | |
config_normal.txt | 4 years ago | |
config_normal_cn_true.txt | 4 years ago | |
config_testing.txt | 4 years ago | |
config_testing_cn_true.txt | 4 years ago | |
config_tumor.txt | 4 years ago | |
config_tumor_cn_true.txt | 4 years ago | |
how_wall_clock_time_split_group_V2.sh.txt | 4 years ago | |
readme.md | 4 years ago | |
set_env.sh | 4 years ago | |
time_all_stats_pred.sh | 4 years ago | |
wall_clock_time_split_group_V2.sh | 4 years ago | |
wall_clock_time_stats_pred.sh | 4 years ago |
1 Image patch extraction
2 Prediction
3 Heatmap stitching
4 Retreiving run-time statistics for prediction
5 Retreiving wall-clock time statistics for extraction and grouping
Run the commands listed in the following subsections to launch Son of Grid Engine (SGE) jobs to extract, group patches in HDF5 files and create a lookup table for every HDF5 file.
qsub ./image_patch_extract/split_main.sh ./config_testing.txt
-- split_main.sh in turn submits split_grp.sh which in turns runs split_grp.py in array job fashion. Every task in the array job processes one slide.
qsub ./image_patch_extract/split_main.sh ./config_normal.txt
qsub ./image_patch_extract/split_main.sh ./config_tumor.txt
The *.sh files mentioned in this section are located under image_patchextract directory while the config*.txt files are at the root directory of the codes.
The lookup tables are created only once and used at Prediction stage for launching array job tasks. These tasks are run in parallel and scalable manner - if there are not enough resourcs for running all tasks then they are queued up automatically and started as resources become available. Each task processes only one group.
The *.sh file mentioned in this section is located under image_patchextract directory while the config*.txt files are located at the root directory of the codes.
The *.sh files mentioned in sections 2.1 and 2.2 below are located under prediction directory while the config_*.txt files are at the root directory of the codes.
After the predictions matrices have been generated an SGE job using heatmap_main.sh SGE scrip could be launched to genertae heatmaps. Two arguments for this launch are: a) type of the slides (test, normal or tumor); b) the root directory of the results, like in below example run:
The files mentioned in this section are located under heatmap_stitch directory.
In time_all_stats_pred.sh file adjust job results root directory, DIR and slides type, PREFIX (normal, test or tumor), like below:
time bash ./time_all_stats_pred.sh
In wall_clock_time_stats_pred.sh file adjust job results root directory, DIR and slides type, PREFIX (normal, test or tumor), like below:
DIR=results_directory
PREFIX=test
Then run:
time bash ./wall_clock_time_stats_pred.sh
Run time_stats_sg_V2.sh Linux scrip with two arguments: a) type of slides (test, normal or tumor) and b) the root directory of the results:
An example run: