Dahlquist:GRNmap Testing

Home        Research        Protocols        Notebook        People        Publications        Courses        Contact       

This page is an electronic notebook to record the results of the GRNmap tests I will peform.

2017-02-16

List of tests that I want to perform:

1. Same input workbook run back to back on the same computer with "clear all" command in between.
2. Same input workbook run back to back on the same computer with no "clear all" command in between.
3. Same input workbook run back to back on the same computer with a computer restart in between.
4. Given the results of the above 3, run the same input workbook 5 times in a row.
5. Same input workbook run on the same machine with no other known programs running vs. lots of other programs running.
6. Same input workbook run on different computers (other lab computers; S120 computers)
7. For the tests that give different results, run the 16 permutations of fixP, fixb, MM/sigmoid, graph/nograph, forward-only/estimate.

Input File

• Downloaded the file 15-genes_28-edges_BK-dHAP4-fam_Sigmoid_estimation.xlsx from the DahlquistLab repository.
• Opened it in Excel on my laptop and verified that no leftover formulas in each worksheet by scrolling through each cell. Left cursor in cell A1 for each worksheet and the "optimization_parameters" worksheet on top.
• Did "Save As" and named it 15-genes_28-edges_BK-dHAP4-fam_Sigmoid_estimation_KD-master-testing-input.xlsx and uploaded to new directory in the repository called kdahlquist_testing_20170216
  • Note that it has just now occurred to me that the issue might be with different versions of MS Excel doing something to the input files.
  • This input file was created with version Microsoft Excel 2013 (15.0.4885.1000) MSO (15.0.4903.1000) 64-bit, Part of Microsoft Office Professional Plus 2013.
    • Information found under File tab > Account (left menu bar) > About Excel button
  • File is 33,467 bytes; will not open or modify it during testing.

GRNmap version

• I might as well use the draft v1.6 version; but since there is not a published release, I will have to use a download of the code from the Master branch; I will be careful to use the same set of files each time and not re-download in case there are changes to the repository. I have already shown that the executable and the code produce the same result on the same computer (back-to-back runs without restarting), so I'm not going to worry about that right now.

Test 1

• Date: 2017-02-16
• Test Performed by: @kdahlquist
• Code Version: Master branch download from 2017-02-16 (v1.6 draft)
• MATLAB Version: 2014b
• Computer on which the model was run: Bio laptop 2 (student account)
• Start: 12:43
• End:
• Purpose: first run to which subsequent runs will be compared
• Issue #323
• First run aborted with Java error; see Issue #323 for details

Test 1 redux

• Installed Java 8 update 121 under Bioadmin account
• Copied code and test file over to this account and started run at 5:01 PM 2017-02-17

Results

• Input sheet: [[Media:]]
• Output sheet: [[Media:]]
• Output .mat file (zipped): [[Media:]]
  • LSE:
  • Penalty term:
  • Number of iterations (counter):
• Figures (all expression graphs .jpg files zipped together): [[Media:]]
• Save the progress figure containing the counts manually: [[Media:]]
• analysis.xlsx containing bar graphs: [[Media:]]
• GRNsight figure of unweighted network: [[Image:]]
• GRNsight figure of weighted network:

  File:*filename here*