-
Notifications
You must be signed in to change notification settings - Fork 1
/
TODO.txt
29 lines (19 loc) · 1.17 KB
/
TODO.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
1. Find a way to make the decision continue splitting: decrease the splitting
variance?? try to make the average maximum depth 4 instead of 3. Maybe try a
larger bagging size?
2. Optimize motif selection. Find more motifs than needed and see how adding
one motif affects the F1 score: optimize how to select the motif score
3. Change the Random Forest model from a classifier to a regression model to predict
the mean toxscore of the group
4. Graph the results from the test data (graph again after the decision tree
receives better results) (DONE)
5. Finish writing manuscript and include all the new information (DONE)
7. Parallelize EVERYTHING! Learn paralelization in Java, or rewrite the motif
extraction algorithm in C++ learn how to use parallelization flags in WEKA
8. Run a peptide motif finding software on the clustered petpides found from
the motif decision tree. LOOK FOR PEPTIDE CLUSTERING SOFTWARES
9. update Gibb's Sampler from 365 to work for peptides and use that as the
motif finder??
10. Compare motifs to known AMPs and see if any match.
11. Learn how to use auto encoder, train it on our data.
12. Look at peptide binding and folding, see if it offers new insights.