Related Genes

[orenbenkiki]

Present some concept of "related genes" to the user which goes beyond "most correlated genes to a base gene". Options are:

• Take the most correlated genes as today, but display a pheatmap of the correlations of these genes with each other (and the original gene). E.g. think of DNTT which might be correlated with both T-cell and B-cell genes, such a view would allow the user to see this more clearly.
• Divide the genes to clusters in advance (similar to the relate_genes function shown in the MC2 vignette), and show the genes in the same cluster as the base gene. E.g., we would end up with "T-cell" clusters and "B-cell" clusters, ideally with no cluster mixing the two.
  The first option is more powerful for "power users". The second option is more convenient for "lazy users". This calls for a true "poduct manager" decision (which type of users do we tailor MCView for? If both, how do we avoid over-complicating the system?).