/Alpha-Frag

Primary LanguagePythonMIT LicenseMIT

Alpha-Frag

Alpha-Frag is a deep neural network to predict the presence of peptide fragment ions. You can use it directly without installation.

Hardware

A GPU with CUDA support

Package

Example

  1. Sequences with amino acid [BJOUXZ] are not supported.
  2. Each C is treated as Cysteine with carbamidomethylation.
  3. Modifications except "M(ox)" are not supported. M(ox) should be written as 'm'.
  4. The length of peptide should be between 7 and 30.
  5. The precursor charge should be between 1 and 4.
  6. Output such as 'y10_2' means 'y' type ions and cleavage size at 10 with charge 2. More than 2 charges fragment ion is not considered.
  • run by case:
python run_by_case.py ACDEFGHIKLMmNPQRSTVWYK 2

output:

ACDEFGHIKLMmNPQRSTVWYK_2: y1_1;y2_1;y3_1;y8_1;y9_1;y10_1;y11_1;y12_1;y13_1;y14_1;y15_1;b2_1;b3_1;b4_1;b7_1;b8_1;b9_1;b10_1
  • run by DataFrame:

The input DataFrame should include 'simple_seq' and 'pr_charge' columns.

see run_by_df.py

Application

One of the application for Alpha-Frag is constructing the qualitative and quantitative scores to improve the identification based on the output of OpenSWATH.

  1. Make a workspace folder containing:
    • *.mzML
    • osw.tsv (the output by OpenSWATH v2.4.0 and named with osw.tsv)
  2. Run: 
    python run_scoring your_workspace_dir