New figure file and updated import_ccle #25
Conversation
aarmey
left a comment
aarmey
left a comment
There was a problem hiding this comment.
One comment, plus remove the built figure files (.png's), and check the build errors.
quinn-mccall
force-pushed
the
quinns-branch
branch
from
aaddb2b to
e53381f
Compare
aarmey
left a comment
aarmey
left a comment
There was a problem hiding this comment.
I'd like to better understand what is necessary to get the matching correlation values. The current none import code is strange. For instance, people always log-transform this type of data.
| # Term 1: y * log(y / mu) = xlogy(y, y) - xlogy(y, mu) | ||
| # xlogy handles y=0 case correctly returning 0. | ||
| term1 = xlogy(y_ij, y_ij) - xlogy(y_ij, mu_ij) | ||
|
|
There was a problem hiding this comment.
I simplified the import code to make it easier to fix issues.
pf2barcode/figures/figure4.py
Outdated
| X = X[X.obs["SW"].isin(good_SW)] | ||
|
|
||
| # Convert matrix to dense for correlation computation | ||
| mat = X.X.toarray() |
There was a problem hiding this comment.
Now converted to dense on import, so this can go.
|
|
||
| results = [] | ||
|
|
||
| for sw, cells in df.groupby("SW"): |
There was a problem hiding this comment.
Please move the calculation of the results to a separate function that only takes in the data it needs for the work. Please add comments to the loops and key operations explaining what is being done.
aarmey
left a comment
aarmey
left a comment
There was a problem hiding this comment.
Check the testing (lining) errors.
No description provided.