If I want to quickly see the top N coordinates associated with a term or how much a term is associated with certain coordinates, I should be able to use one of these methods.

sometimes mixture of gaussians returns nans

• choose gaussian sd
• simple group mean of radius r

Do we want this to be a necessary dependency for just using NSABA locally?

When quant =None, analysis gets hung on threshold step.

make a txt file of NS terms that will be mostly empty.
Each time you run a term try loading using the preinitialized txt file.
If term isn't run yet, run it and save the vector to the txt file.

to accommodate requirement for N.ge['entrez_id']['GE']

allow ge_ns_mat to take in a lists of terms and/or gene entrez ids so we can do multiple linear regression easily.

generalize estimation methods

id_dict only has 7k registered IDs, ~11k are registered.

Different methods produce p values to say how well our estimates were:
methods:
term_to_genes(one_term, some_genes,alpha)
returns how many genes were in the alpha quartile

gene_to_terms(one_gene, some_terms,alpha)
returns how many terms were in the alpha quartile

...more coming

i.e. using certain coords to estimate ge returns weird heat maps