search

glasgowcompbio / PALS

Ranking metabolite (and other omics sets) by their activity levels via SVD
https://pals.glasgowcompbio.org
MIT License
2 stars 1 forks source link

Activity Level Calculation #32

Closed joewandy closed 4 years ago

joewandy commented 4 years ago

Just wondering if activity levels are calculated correctly ..

  1. At this point, measurement_df has already been standardized (0 mean and unit variance). Should it be divided by np.sqrt(len(row_ids)) before the SVD?

    pathway_data = measurement_df.loc[row_ids] / np.sqrt(len(row_ids))  # DF selected from peak IDs.

    https://github.com/glasgowcompbio/PALS/blob/master/pals/PLAGE.py#L276

  2. PLAGE paper defines

    the activity level of a pathway in a given sample j is taken as the coefficient _cj for the first metagene.'

This is c[0] in our code, but below it is also multiplied by the corresponding eigenvalue d[0]. Is this correct?

pw_act_list.extend(list(c[0] * d[0]))

https://github.com/glasgowcompbio/PALS/blob/master/pals/PLAGE.py#L283

RonanDaly commented 4 years ago

Makes no change to results, but removed from code to make things clearer. Commit 118199150ce342f6c9c115a91a937bb32df5d7eb.