DeepMRM

DeepMRM is a targeted proteomics data interpretation package using deep-learning. It takes MRM, PRM, or Data-Independent Acquisition (DIA) data and target list as input and outputs peaks of targeted peptides, along with their abundances and quality scores. The targeted endogenous (light) peptides are detected and quantified using their stable isotope labeled (heavy) peptides.

Installation (without GPUs)

You can use DeepMRM with conda environment or docker.

Conda environment

We recommend to run DeepMRM in a conda environment such that your environment for DeepMRM and dependencies are separate from other Python environments.

Download and install Anaconda
You can download Anaconda from https://www.anaconda.com/download
Create a conda environment
Open the terminal (Linux or MacOS) or the Anaconda Prompt (Windows).
Create a new conda environment, named deepmrm, and activate it by running following commands:
```
conda create -n deepmrm python=3.9
conda activate deepmrm
```
Clone the repository and move to the deep-mrm directory in the terminal.
```
git clone https://github.com/bertis-informatics/deep-mrm.git
```
Install required packages
```
pip install -r requirments.txt
```
Install DeepMRM package
```
conda develop .
```
This will output the deep-mrm directory path.

Docker

Download and install docker desktop (https://www.docker.com/get-started/)
Pull DeepMRM docker image
```
docker pull jungkap/deepmrm:v1.6
```

How to run

Conda environment

Open the terminal (Linux or MacOS) or the Anaconda Prompt (Windows), and activate your deepmrm conda environment created during the installation

Run DeepMRM running script with a mass-spec file (mzML) and a target list (csv).
You can find sample input files in sample_data directory.
For example, if you are in deep-mrm directory and want to run DeepMRM for sample data, then run the following command:

python deepmrm/predict/make_prediction.py \
           -target ./sample_data/sample_target_list.csv \
           -input ./sample_data/sample_mrm_data.mzML

Syntax

-input: string path for input mzml file

-target: string path for target csv file. The csv file should contain following columns:

peptide_id: unique ID for each targeted peptide (heavy and light transitions are grouped by peptide_id)
precursor_mz: m/z values for targeted precursor ions
product_mz: m/z values for targeted fragment ions

is_heavy: flag indicating light or heavy peptides (TRUE or FALSE)

Example of target list

peptide_id	precursor_mz	product_mz	is_heavy
PDAC0008	798.36005	966.4759	FALSE
PDAC0008	798.36005	1210.596	FALSE
PDAC0008	798.36005	1472.6759	FALSE
PDAC0008	800.36676	972.496	TRUE
PDAC0008	800.36676	1216.596	TRUE
PDAC0008	800.36676	1478.696	TRUE

-tolerance: int, default=10. Ion match tolerance (in PPM). This is ignored in the case of MRM data.

Docker

You can run DeepMRM through docker with following command. You should specify {local_dir_for_input_data}, {target_csv_file_name}, and {mzml_file_name} before running.

docker run -v {local_dir_for_input_data}:/input_data jungkap/deepmrm \
            -target /input_data/{target_csv_file_name} \
            -input /input_data/{mzml_file_name}

For example, you can run DeepMRM for sample data in the deepmrm directory as follows.
In Windows, you need to replace forward slash with backslash in the local directory path.

docker run -v ./sample_data:/input_data jungkap/deepmrm:v1.6 -target /input_data/sample_target_list.csv -input /input_data/sample_mrm_data.mzML

License

DeepMRM package is freely available for academic research, non-profit or educational purposes only under DeepMRM-LICENSE. For commercial use of DeepMRM, please contact deepmrm@bertis.com.

bertis-informatics/deep-mrm

DeepMRM

Installation (without GPUs)

Conda environment

Docker

How to run

Conda environment

Docker

License