How TDGC works

Get started

After installing TDGC, the user is able for performing the process.

TDGC's workflow is divided in two main steps: Training step, which some good and bad spectra are used to teach the algorithm to dissect the good of the bad ones; and Prediction step, which according to classifier model created in the training step, the algorithm will classify all spectra from a dataset.

Follow step-by-step below to load the raw files and set up the parameters to extract the high-quality ms/ms spectra.

  1. In the field 'Noisy Examples', select the ms/ms spectra, which represent the poor set for training the algorithm.*
  2. In the field 'Positive Examples', select the ms/ms spectra, which represent the good set for training the algorithm.*
  3. Select the bin size of the ms/ms spectrum. The default value is 25.
  4. Select the m/z range that the algorithm will consider to train all spectra for creating the library.
  5. Click on 'Train Ratio' button.

*TDGC has already stored a training set for ms/ms spectra obtained from the EThcD fragmentation method. In addition, in the Download tab, it's possible to download a training dataset from HCD fragmentation method.

  1. On 'Classify Raw file in directory' tab, select the directory which contains all RAW files to be processed.
  2. On 'Spectra classified as noisy (0) with # or more envelopes should be considered' means that some spectra could be classified as noisy, however they have enough isotopic envelopes to be considered as good. Thus, set a number of minimum envelopes. The default value is 20.
  3. Other wise, on 'Spectra classified as good (1) with # or less envelopes should be considered as noise' means that some spectra could be classified as good, however they don't have enough isotopic envelopes. Thus, set a number of maximum envelopes. The default value is 5.
  4. Click on 'Go' button.

The results will be generated in the same directory of the input data.