/SPA-STOCSY

Primary LanguageHTML

SPA-STOCSY

SPA_STOCSY is an R pipeline for the automatic identification of both "known" and "unknown" metabolites in given samples.
With given input samples, SPA_STOCSY offers automatic algorithms to estimate all the parameters and generate the final results.
Here is an example of using SPA_STOCSY to analyze 10 NMR samples from brain homogenates of Drosophila melanogaster.

To apply SPA_STOCSY for NMR spectra with diversified resolutions:

  • For high-resolution NMR spectra, we provide the low-dimension SPA-STOCSY tutorial. The metabolite clusters from SPA are represented by their average peak intensities, which is the input for the STOCSY step.

Low-dimension SPA-STOCSY tutorial

  • For low-resolution NMR spectra, we provide the high-dimension SPA-STOCSY tutorial. For each metabolite cluster within SPA, all the local maximums are used as input for the STOCSY step, thus each cluster is represented by a list of peaks for correlation calculation, instead of the average value only.

High-dimension SPA-STOCSY tutorial

Data

Some data used in the paper are stored in the input_data/folder.

  • 10_meta_simulation.csv: 50 synthesized samples from 10 metabolites
  • htt_c12.csv: 10 NMR samples from brain homogenates of Drosophila melanogaster
  • human_cell_data.csv: 22 NMR samples from human cells
  • ref_final_lib.csv: reference library for 257 metabolites with their cluster information
  • chx_meta_with_single_clus.csv: metabolite name list used for singlet filtering

Installation

The main functions in SPA_STOCSY are stored in the source_codes/ folder.
To use SPA_STOCSY for spectral analysis:

  1. The user can download the SPA-STOCSY repository to their local directory
  2. Follow the tutorial to load the source functions for SPA-STOCSY
  3. Then the user can use the data to explore SPA-STOCSY following the tutorial

Input

  • The input file takes multiple NMR samples as columns. The first column should be specified as a list of chemical shift values.
  • The NMR samples are highly recommended to get normalized on integral.

Output

  • The output location is defined by the user before running the tool as in the example
  • The output files contain:
    • Multiple visualization plots during the running process
    • 'SPA_cluster_peaks.csv': peaks detected in each SPA cluster
    • 'STOCSY_corr_clusters.csv': correlated SPA clusters generated from STOCSY
    • 'auto_identify_metabolites.csv': list of identified metabolites with a set detection rate