search

KrishnaswamyLab / MAGIC

MAGIC (Markov Affinity-based Graph Imputation of Cells), is a method for imputing missing values restoring structure of large biological datasets.
GNU General Public License v2.0
347 stars 96 forks source link

Magic + seurat workflow #204

Open anemartinezlarrinaga2898 opened 2 years ago

anemartinezlarrinaga2898 commented 2 years ago

Hello!

I have found the paper and I think is pretty nice tool, but have some questions:

  1. Can It be applied with integrated data? The data has been integrated with Seurat and normalized with log.
  2. Which one will be the best normalization method? SCT or stay with the log normalization.
  3. Also does anyone perform downstream analysis with the default assay being the MAGIC output? Such as ligan receptor analysis

Thank you in advanced!

LucaTucciarone commented 2 years ago

Wow, we have the same questions and applications. Did you solve it?

anemartinezlarrinaga2898 commented 2 years ago

Nop, I decided to look for other tools where I have more info on how to work with them!

Sorry

LucaTucciarone commented 2 years ago

Could u share what tools you landed on in the end? I tried their normalization but I get a ton of NaNs, I am running using SCT now, I shall see what happens but I am open to other tools

anemartinezlarrinaga2898 commented 2 years ago

With Magic I manage to run it but in Python, couldn't manage to do it in R. Also I dint like that the parameter selection is so biased to the user.

Im going to use SAVER https://www.nature.com/articles/s41592-018-0033-z it does benchmark with Magic and it seems more robust and less biased.

Hope it helps!