Normalization

[MicTott]

Roadmap for normalization and Dimensionality Reduction

Step 1: Subset genes to only 1:1 orthologs

• Convert baboon and macaque genes to human while retaining only 1:1 orthologs using orthogene::convert_orthologs with non121_strategy="drop_both_species"
• Get common NHP orthologs, Drop orthologs not in human dataset to get final list of common orthologs
• Subset human, baboon, and macaque SCEs to only common 1:1 orthologs

Step 2: Log normalization and scale factos across species

Within each species ...

• Deal with compositional biases by pre-clustering using quickCluster. This avoids the assumption that most genes are non-DE across the population
• Calculate scaling factors using computeSumFactors
• Log normalize the count data with 'logNormCounts'