Appendix

Project folder structure guidance

The guidances notes provide an overview of the folder structure used in this handbook

• All projects live in an S3 bucket

• The directory structure is always <bucket-name>/projects/<project_name>

• In the <project_name> folder there are sub directories

  • The first subdirectory is workspace

  • Other subdirectories are batches of data

• The batches of data are labeled by date and include images and illum folders

  • In the images folder there exist different plates storing raw image data

  • The illum folder is identical to the images folder in terms of structure

    • illum is an output of the first stage of CellProfiler pipeline that stores a function to adjust the plates in images

• workspace also has subdirectories

  • analysis - includes subfolders mirroring the Batch nesting

    • Within each batch folder, the CellProfiler results are stored in plate_id

      • Within each plate folder there is an analysis folder

        • Inside this analysis folder, each well has its own folder (e.g. A01-1)

          • A and 01 refer to the row and column of the plate, 1 refer to sites per well

          • If the grouping was done by well instead of by site, this would be A01, without the suffix of -1

        • Note that this analysis folder is customizable

          • There are typically 384 (# of wells) x 9 (# of sites per well) subfolders

            • 384 well plate

            • 9 different pictures

        • Within the site folder (e.g. A01-1) there are five csv files

          • Cells.csv

            • Each row are measurements of one cell

          • Cytoplasm.csv

            • Another object similar to Cells.csv

          • Nuclei.csv

            • Another object similar to Cells.csv

            • These three object files can be concatenated by column Objects.csv

          • Experiment.csv

            • Stores metadata for the CellProfiler run, including the CellProfiler pipeline itself

          • Image.csv

  • backend - also includes batch nesting

    • batch nesting

      • plate nesting - stores summaries of each plate (all .csv files also have .gct formats (for input into Morpheus)

        • <plate_id>.sqlite - inner join of all objects in a well, and then stacked (so all data for each well in a single plate)

        • <plate_id>.csv - per well means for each well on the plate

        • <plate_id>.augmented.csv - same as .csv except it includes the metadata

        • <plate_id>._normalized.csv - some z scored version of augmented

        • <plate_id>._normalized_variable_selected.csv - across all the plates in the batch

          • Three feature selection steps

            • Variance threshold

            • Correlation threshold (decorrelate feature set)

            • Replicate correlation filter (>0.6)

  • parameters - same structure as backend but with metadata results (e.g. the features selected in variable selection)

  • software

    • This is where the project’s github repository lives.

    • The scripts in the handbook assume that this be named as the same name as the Project folder. To rename it, pay careful attention to paths when executing the commands in the handbook.

Directory structure

├── 2016_04_01_a549_48hr_batch1
│   ├── illum
│   │   └── SQ00015167
│   │       ├── SQ00015167_IllumAGP.npy
│   │       ├── SQ00015167_IllumDNA.npy
│   │       ├── SQ00015167_IllumER.npy
│   │       ├── SQ00015167_IllumMito.npy
│   │       └── SQ00015167_IllumRNA.npy
│   └── images
│       └── SQ00015167__2016-04-21T03_34_00-Measurement1
│         ├── Assaylayout
│         ├── FFC_Profile
│         └── Images
│             ├── r01c01f01p01-ch1sk1fk1fl1.tiff
│             ├── r01c01f01p01-ch2sk1fk1fl1.tiff
│             ├── r01c01f01p01-ch3sk1fk1fl1.tiff
│             ├── r01c01f01p01-ch4sk1fk1fl1.tiff
│             └── r01c01f01p01-ch5sk1fk1fl1.tiff
└── workspace
    ├── audit
    │    └── 2016_04_01_a549_48hr_batch1
    ├── analysis
    │    └── 2016_04_01_a549_48hr_batch1
    │        └── SQ00015167
    │            └── analysis
    │                └── A01-1
    │                    ├── Cells.csv
    │                    ├── Cytoplasm.csv
    │                    ├── Experiment.csv
    │                    ├── Image.csv
    │                    ├── Nuclei.csv
    │                    └── outlines
    │                        └── SQ00015167
    │                            ├── A01_s1--cell_outlines.png
    │                            └── A01_s1--nuclei_outlines.png
    ├── backend
    │   └── 2016_04_01_a549_48hr_batch1
    │       └── SQ00015167
    │           ├── SQ00015167.csv
    │           └── SQ00015167.sqlite
    ├── images
    │   └── 2016_04_01_a549_48hr_batch1 -> /home/ubuntu/bucket/projects/2015_10_05_DrugRepurposing_AravindSubramanian_GolubLab_Broad/2016_04_01_a549_48hr_batch1/images/
    ├── load_data_csv
    │   └── 2016_04_01_a549_48hr_batch1
    │       └── SQ00015167
    │           ├── load_data.csv
    │           └── load_data_with_illum.csv
    ├── log 
    │   ├── create_csv_from_xml
    │   └── collate       
    ├── metadata
    │   └── 2016_04_01_a549_48hr_batch1
    │       ├── barcode_platemap.csv
    │       └── platemap
    │           └── C-7161-01-LM6-006.txt
    ├── pipelines
    ├── status
    └── software
        ├── Distributed-CellProfiler
        └── pe2loaddata