A deep learning method for Human Plasma Proteome identification and quantification.

Download and install Anaconda.

Create a new environment with Python 3.9+:

conda create --name yourEnv python=3.9

Activate the environment:

conda activate yourEnv

If you have a GPU, you can install CUDA. If not, skip this step.

Visit the CUDA Toolkit Archive.

Select your computer's architecture.

Choose CUDA 11.7.0 and download the toolkit.

Install the necessary packages from the requirements.txt file:

conda install --yes --file requirements.txt

Create a folder to store your raw data. Here is an example of how your folder structure should look:

ttp_20230702_CCS
├── ttp_20230702_CCS_Plasma_trypsin_Mix1_20min_system-test-1_Slot1-10_1_12901.d
├── ttp_20230702_CCS_Plasma_trypsin_Mix1_20min_system-test-2_Slot1-10_1_12902.d
├── ttp_20230702_CCS_Plasma_trypsin_Mix1_20min_system-test-3_Slot1-10_1_12903.d
├── ttp_20230702_CCS_Plasma_trypsin_Mix1_20min_system-test-4_Slot1-10_1_12904.d
└── ttp_20230702_CCS_Plasma_trypsin_Mix1_20min_system-test-5_Slot1-10_1_12905.d

Create a folder to store the raw library files with the .tsv suffix.

Run library_generation.py to generate the target and decoy libraries with the .npy suffix.

In the configs folder, you will find two JSON files:

• preprocess_configs.json
• model_configs.json

You can modify the following fields:

• library: Replace target_path and decoy_path with your library paths.
• save_root_path: Set the path to save the preprocessed data.
• mobilityDistanceThreshold: Set to 0.05 or 0.1 to filter MS2 data.

You can modify the batch_size field.

Run the main.py script:

python main.py --data_path='raw_data_path' --num_process=5

The results, including the peptide and protein group quantification in TSV format, will be saved in the save_root_path folder.