From SnOwy - Ed's Wiki Notebook

20100520 To Do

• Deadline on May 22
• Methods done, need to put in inline citations
• Background done, need to put in inline citations
• Results at 0.00-- postpone until tomorrow (need to prepare a group meeting)

20100517 To Do

• Boils down to three things
  1. Methods
     • Finishing neural network math ...
  2. Background
  3. Results

20100513 To Do

• Proceeding onto NGN feed forward and back propagation math.
• Add a reference for the isopentanol Monet rendering.
• Add and discuss all results ever generated in review style.
• Complete QSAR intro in intro
• Complete SMILES and InChI intro in methods
• Completed NGN architecture math
• Replaced previous monet molecule with a new monet molecule.

20100510 To Do

• No more experiments, reading/reviewing/writing methods...

20100504 To Do

• Got Makefile from Chris... testing...
• Confirmed, does not work correctly with present grammar-- get grammar file from Pewter and test with that.

20100430 To Do

• Write core of paper-- finish the methods first, I think this will resemble a methodology report
• Don't put too much worry into the InChI stuff yet, we'll focus on this when Chris' makefile is available
• Don't worry about non-parametric statistics yet (recall: this was an alternative proposed to describe Chris' results)

20100430 Briefings

Dear Chris and Dr. Kremer,


Dr. Kremer's been added to this conversation.

Dr. Kremer, we're discussing (1) the BMC2010 paper due May 15, (2)
that authorship should include Chris because we'll be revisiting his
results and (3) the possibility of getting Chris to do InChI-NGN on
his datasets soon.

Chris, The executable *should not* be recognized :P It isn't a windows
binary afterall; yes, I will wait :D

When this is done, would you have time to run your datasets on
SharcNet? Just repeat the exact same procedure as with SMILES but with
InChI instead this time.

Because we will be discussing all of the NGN results including those
pertaining to the experiments you ran, you would be on the authorship
list.

(And of course, if you're actually running experiments-- that would
only be fair.)

Dr. Kremer and Chris,

It looks like there's 14 days to prepare the paper ... making the math
palatable for BMC is a challenge (this journal is very biology heavy).

I have started by porting an older project in Stefan's repository to
this location:
projects/Ma/2010Paper/

The makefile is here:
projects/Ma/2010Paper/ngnmath/

Chris, don't worry about the paper writing-- I will ask you if there
are interpretation problems with the previous CIBCB2010 content.


Thanks,
Ed

Paper

• fragment version needs to be worked into a draft version
• now reviewing fragment version

InChI Parsing

• when my binary is run natively, it'll parse Chris' dataset
• when my binary is wrapped within Chris' script, parsing the dataset's inchis results in a parse error
• recovered my "fixit" directory! It was in the repository all along.

Briefings

Hey Eddie,
 The scripts are:
  sort.py - parses the inchi csv files and separates them into the animal, animal & organ, and no sort databases
  converge.py - tries to converge the given data set
  group and rand train.py - running training and testing programs
  datacollect.py - is for my statistical data gathering
  stats.py - runs the stats equations
Hope that helps,
Chris