From c5b51d867bfd86bc253d8497dafd72f569f96e0f Mon Sep 17 00:00:00 2001 From: Niranj Chandrasekaran <niranjchandrasekaran@users.noreply.github.com> Date: Fri, 23 Feb 2024 12:18:13 -0500 Subject: [PATCH] Update README.md Fix the name of the repo --- README.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index 41f199b..d0addce 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ <small><i><a href='http://ecotrust-canada.github.io/markdown-toc/'>Table of contents generated with markdown-toc</a></i></small> -Image-based features from the cell images were extracted using [CellProfiler](https://cellprofiler.org/) and assembled as single cell profiles, which were aggregated, annotated, normalized and feature selected using [pycytominer](https://github.com/cytomining/pycytominer). Image-based features were also extracted using [DeepProfiler](https://github.com/cytomining/DeepProfiler) which were annotated and [spherized](https://en.wikipedia.org/wiki/Whitening_transformation). The resulting profiles were analyzed using the [notebooks in this repo](https://github.com/jump-cellpainting/2023_Chandrasekaran_submitted/tree/main/benchmark). Steps for reproducing the data in this repository are outlined below. +Image-based features from the cell images were extracted using [CellProfiler](https://cellprofiler.org/) and assembled as single cell profiles, which were aggregated, annotated, normalized and feature selected using [pycytominer](https://github.com/cytomining/pycytominer). Image-based features were also extracted using [DeepProfiler](https://github.com/cytomining/DeepProfiler) which were annotated and [spherized](https://en.wikipedia.org/wiki/Whitening_transformation). The resulting profiles were analyzed using the [notebooks in this repo](https://github.com/jump-cellpainting/2024_Chandrasekaran_NatureMethods/tree/main/benchmark). Steps for reproducing the data in this repository are outlined below. # Step 1: Download cell images @@ -111,8 +111,8 @@ After generating the well-level CellProfiler-based features, use Pycytominer to To regenerate all the profiles, clone this repo, download the files and activate the conda environment. Before issuing the following commands, Install [Miniconda](https://docs.conda.io/en/latest/miniconda.html). ```bash -git clone https://github.com/jump-cellpainting/2023_Chandrasekaran_submitted -cd 2023_Chandrasekaran_submitted +git clone https://github.com/jump-cellpainting/2024_Chandrasekaran_NatureMethods +cd 2024_Chandrasekaran_NatureMethods git lfs pull git submodule update --init --recursive conda env create --force --file environment.yml @@ -136,7 +136,7 @@ This creates the profiles in the `profiles/` folder for all the plates in each b | `<plate_ID>_normalized_feature_select_plate.csv.gz` | Feature selected normalized to whole plate profiles | | `<plate_ID>_normalized_feature_select_negcon_plate.csv.gz` | Feature selected normalized to negative control profiles | -Annotated DeepProfiler profiles are spherized using [this notebook](https://github.com/jump-cellpainting/2023_Chandrasekaran_submitted/blob/main/benchmark/old_notebooks/3.spherize_profiles.ipynb). +Annotated DeepProfiler profiles are spherized using [this notebook](https://github.com/jump-cellpainting/2024_Chandrasekaran_NatureMethods/blob/main/benchmark/old_notebooks/3.spherize_profiles.ipynb). # Step 4: Run the benchmark script The benchmark scripts compute `Average Precision (AP)` for various retrieval tasks, such as, retrieving replicates against negative controls, retrieving perturbation pairs against non-pairs, and retrieving gene-compound pairs against non-pairs. `AP` was calculated using the `Feature selected normalized to negative control profiles` (well-level profiles).