Please reply with any questions you had from Nicky’s presentation on 3/22.
cmi_pb_240322.pdf (1.1 MB)
Thank you for your nice presentation Nicky! I was wondering if you did any imputation for MOFA?
Hi Lisa. Thanks for the question. I didn’t do any imputation.
MOFA can handle missing data, so imputation wasn’t necessary. It was still able to calculate factors for all samples, even if those samples had missing values.
The only downside was that we took a combination of MOFA factors and baseline data as predictors. For the baseline data we needed to impute data anyway, so in the end we had to throw away samples. In future I would probably impute anyway, just to be sure we can keep all the baseline data.