amfinder:AMFinder套件包含Python脚本，该脚本可从墨迹斑斑的植物根部图片中对菌根结构进行高通量注释，以及用于...

# AMFinder The Automatic Mycorrhiza Finder (AMFinder) allows for automatic computer vision-based identification and quantification of AM fungal colonisation and intraradical hyphal structures on ink-stained root images using convolutional neural networks. ## Summary 1. [Command-line script (`amf`)](#amf) 2. [Standalone interface (`amfbrowser`)](#amfbrowser) ## Command-line script (`amf`)<a name="amf"></a> ### Installation instructions `amf` requires at least **Python 3.6**. Below is a typical installation, using a virtual environment to install dependencies: ```bash $ python3.7 -m venv amfenv $ source amfenv/bin/activate (amfenv) $ python -m pip install -r requirements.txt (amfenv) $ deactivate ``` **Note:** `amf` is not compatible with the latest Keras/Tensorflow versions due to [a bug affecting class weights](https://github.com/tensorflow/tensorflow/issues/40457). ### Prediction mode Predict fungal colonisation or intraradical hyphal structures within plant root images. ``` $ amf predict <options> <jpeg/tiff images> ``` Available `<options>` are listed below: |Short|Long|Description|Default| |-|-|-|-| |`-h`|`--help`|Display this help.| |`-t`|`--tile_size`| Tile size, in pixels.|126| |`-net`|`--network`|Select a pre-trained network in `trained_networks`.|None| Pre-trained networks are available in folder [`trained_networks`](amf/trained_networks): |File name|Annotation level|Description| |-|-|-| |[CNN1v1.h5](amf/trained_networks/CNN1v1.h5)</a>|Stage 1|Ink-stained, ClearSee-treated root pictures (flatbed scanner/microscope).| |[CNN1v2.h5](amf/trained_networks/CNN1v2.h5)|Stage 1|Same, but hue- and saturation-insensitive.| |[CNN2v1.h5](amf/trained_networks/CNN2v1.h5)|Stage 2|Ink-stained, ClearSee-treated microscope root pictures.| ### Training mode Train `amf` on a different image dataset, or refine existing models. ``` $ amf train <options> <jpeg/tiff images> ``` Available `<options>` are listed below: |Short|Long|Description|Default| |-|-|-|-| |`-h`|`--help`|Display this help.| |`-t`|`--tile_size`| Tile size, in pixels.|126| |`-b`|`batch_size`|Training batch size.|32| |`-k`|`--keep_background`|Do not skip any background tile.|False| |`-a`|`--data_augmentation`|Activate data augmentation.|False| |`-s`|`--summary`|Save CNN architecture and graph.|False| |`-o`|`--outdir`|Folder where to save trained model and CNN architecture.|cwd| |`-e`|`--epochs`|Number of training cycles.|100| |`-p`|`--patience`|Number of epochs to wait before early stopping.|12| |`-lr`|`--learning_rate`|Learning rate used by the Adam optimiser.|0.001| |`-vf`|`--validation_fraction`|Fraction of tiles used as validation set.|15%| |`-1`|`--CNN1`|Train for root colonisation.|True| |`-2`|`--CNN2`|Train for intraradical hyphal structures.|False| |`-net`|`--network`|Name of the pre-trained network.|None| For large datasets, running the script on a high-performance computing (HPC) equipment is recommended. An example using [Slurm](https://slurm.schedmd.com/) workload manager is provided below. ```bash #! /bin/bash #SBATCH -e train.err #SBATCH -o train.out #SBATCH --mem=10G #SBATCH -n 10 BASE=/home/user/amf source $BASE/amfenv/bin/activate $BASE/amf train $BASE/training_data/*jpg deactivate ``` ### Diagnostic mode Use the following command to determine **precision** and **specificity** of a trained network: ```bash $ amf diagnose -net <trained_network> <images> ``` ## Standalone interface (`amfbrowser`)<a name="amfbrowser"></a> The standalone interface `amfbrowser` allows to browse, amend and validate `amf` predictions. Installation instructions are detailed below for the main platforms.  ### Installation instructions<a name="amfbrowseronlinux"></a> #### Linux 1. Download and install the OCaml package manager [OPAM](https://opam.ocaml.org/doc/Install.html). 2. Using [`opam switch`](https://opam.ocaml.org/doc/Usage.html#opam-switch), install **OCaml 4.08.0** (older versions won't work). 3. Install `amfbrowser` dependencies: ```bash $ opam install dune odoc lablgtk cairo2 cairo2-gtk magic-mime camlzip ``` You may be required to install development packages, including `libgtk2.0-dev` and `libgtksourceview2.0-dev`. 4. Retrieve `amfbrowser` sources and build: ``` $ git clone [email protected]:SchornacklabSLCU/amfinder.git $ cd amfinder/amfbrowser $ ./build.sh ``` 5. Copy the folder `data` to your local application folder. A typical path would be `~/.local/share/amfinder/data`. 6. The binary `amfbrowser.exe` is ready to use. #### MacOS Same as Linux, but you will need [Homebrew](https://brew.sh/index_fr) to install packages. #### Windows 10 `amfbrowser` can be installed and run on Windows 10 after activation of the Windows Subsystem for Linux (WSL). WSL runs a a GNU/Linux environment directly on Windows, unmodified. **Admin rights are required to activate WSL**. 1. Activate the [Windows Subsystem for Linux (WSL)](https://docs.microsoft.com/en-us/windows/wsl/install-win10). Then, go to Windows App store and install a Linux distribution (recommended distributions are [Ubuntu](https://ubuntu.com/) and [Debian](https://www.debian.org/index.html), but many others should work too). 2. Install an OCaml build system based on the `brew` package manager: ```bash $ sudo apt update $ sudo apt upgrade $ sudo apt autoclean $ sudo apt install curl build-essential git $ sh -c "$(curl -fsSL https://raw.githubusercontent.com/Linuxbrew/install/master/install.sh)" $ test -d ~/.linuxbrew && eval $(~/.linuxbrew/bin/brew shellenv) $ test -d /home/linuxbrew/.linuxbrew && eval $(/home/linuxbrew/.linuxbrew/bin/brew shellenv) $ test -r ~/.bash_profile && echo "eval \$($(brew --prefix)/bin/brew shellenv)" >> ~/.bash_profile $ echo "eval \$($(brew --prefix)/bin/brew shellenv)" >> ~/.profile $ brew install gpatch opam gtk+ cairo ``` 3. Follow the [Linux installation instructions](#amfbrowseronlinux). Please note that sandboxing does not work on Windows. OPAM should be initialized using `opam init --disable-sandboxing`. 4. Install a X server (for instance, [Xming](https://sourceforge.net/projects/xming/)), then configure bash to tell GUIs to use the local X server. For instance, use `echo "export DISPLAY=localhost:0.0" >> ~/.bashrc`. Detailed instructions are available on the internet.