I noticed that the MAGIC analysis does not include code to perform two preprocessing steps recommended by the MAGIC authors:
Library size normalization (scaling)
square-root transformation
These steps don't feature in the quickstart section, but are present in both tutorials linked on the author's github repo.
If these were omitted in the analysis featured here, could that explain the high correlation of MAGIC gene expression scores and the (log) of the total counts you show in Figure 4A/B?
The MAGIC authors explain the need for library size normalization e.g. in this tutorial
After filtering, the next steps are to perform library size normalization and transformation. Log transformation is frequently used for single-cell RNA-seq, however, this requires the addition of a pseudocount to avoid infinite values at zero. We instead use a square root transform, which has similar properties to the log transform but has no problem with zeros.
I noticed that the MAGIC analysis does not include code to perform two preprocessing steps recommended by the MAGIC authors:
These steps don't feature in the
quickstartsection, but are present in both tutorials linked on the author's github repo.If these were omitted in the analysis featured here, could that explain the high correlation of MAGIC gene expression scores and the (log) of the total counts you show in Figure 4A/B?
The MAGIC authors explain the need for library size normalization e.g. in this tutorial