settings

[dgcovell]

I am having difficulties when specifying settings from within python. I would like to have the following: gdsc2.settings {'regression_alpha': 0.01, 'pvalue_threshold': 0.1, 'savefig': 'True', 'feature_factor_threshold': 3, 'analysis_type': 'PANCAN', 'animate': True, 'include_MSI_factor': True, 'version': '0.17.1', 'MSI_factor_threshold': 2, 'regression_method': 'Lasso', 'include_media_factor': False, 'FDR_threshold': 75, 'pvalue_correction_method': 'qvalue', 'volcano_FDR_interpolation': True, 'fontsize': 25, 'pvalue_correction_level': 'global', 'volcano_additional_FDR_lines': [0.01, 0.1, 10], 'equal_var_ttest': True, 'regression_L1_wt': 0.5, 'regression_formula': 'auto', 'minimum_nonna_ic50': 6, 'effect_threshold': 0, 'directory': '/home/covelld/PYTHON/gdsctools-0.17.0/data/python_htmp_gdsc_anova'}

So far, I do not find the specified 'directory', and I am not getting any figures. I am running python.2.7.0 under a bash shell and have specified DISPLAY = localhost:0:0.

I have also tried the standalone version as follows: gdsctools_anova --input-ic50 IC50_v17.csv --drug 1 --read-settings settings_test.json --input-drug-decode test_drug_decode_erlotinib.csv Welcome to GDSCTools standalone

TISSUE FACTOR : included MEDIA FACTOR : NOT included MSI FACTOR : included FEATURE FACTOR : included ODAF mode [-----------------100%-----------------] 47 of 47 complete in 0.5 sec Found 41 possible associations

Creating all figure and html documents in html_gdsc_anova Creating main HTML page in directory html_gdsc_anova openning html_gdsc_anova/index.html Creating individual HTML pages for each drug [-----------------100%-----------------] 1 of 1 complete in 0.1 sec

Creating individual HTML pages for each feature

Creating individual HTML pages for each significant association

where settings_test.json contains {"regression_alpha": 0.01, "pvalue_threshold": 0.01, "savefig": true, "feature_factor_threshold": 3, "analysis_type": "PAN CAN", "animate": true, "include_MSI_factor": true, "version": "0.17.1", "MSI_factor_threshold": 2, "regression_method": "L asso", "include_media_factor": false, "FDR_threshold": 25, "pvalue_correction_method": "holm-sidak", "volcano_FDR_interpol ation": true, "fontsize": 25, "pvalue_correction_level": "global", "volcano_additional_FDR_lines": [0.01, 0.1, 10], "equal _var_ttest": true, "regression_L1_wt": 0.5, "regression_formula": "auto", "minimum_nonna_ic50": 6, "effect_threshold": 0, "directory": "/home/covelld/PYTHON/gdsctools-0.17.0/gdsctools/data/standalone_html_gdsc_anova"}

Any help would be most appreciated.

Thanks, David

[cokelaer]

Hi David, I do not know off hand. In principle, the data should be in "/home/covelld/PYTHON/gdsctools-0.17.0/gdsctools/data/standalone_html_gdsc_anova" In which directory did you run the analysis ?

[dgcovell]

Hi Thomas,

Thanks for the reply. Hopefully you met your deadline. In the meantime, I have done my best to improve my python literacy. However, I am still learning, as well as learning the syntax for gdsctools. I appreciate you tolerance in addressing my requests. My goal is to be able to reproduce existing results and provide additional content for completing my analysis. If possible let me provide areas that need your help. Recall that I am attempting to complete an ANOVA to reproduce what is on the web, and what is in TableS4C from the June 2016 paper. I have chosen the results for erlotinib as a test case (row 286 in TableS4C), using NOTCH1_mut and PANCAN, and am running the same calls from python and standalone gdsc.

Here we go:

1. My systems folks have yet to solve what appears to be an X11 forwarding issue. The typical ssh -X username@machine does not solve the error message of “no display name and no $DISPLAY environment variable”. This means that a cannot generate any plots as illustrated in the ODAF section 4.3.2 of the write-up for ANOVA analysis (introduction). Any suggestions would be helpful. Apparently the installation was done using repo. Should I reinstall using conda?

2. The python call: gdsc = ANOVA(ic,gf); where ic=IC50(“IC50_v17.csv”) and gf=GenomicFeatures(“genomic_features_v17.csv”) runs. I have not yet figured out i) how to parse gdsc for content, vars(gdsc) generates too much output, dir(gdsc) give the content in gdsc and print(gdsc.settings) prints, for example, the settings. As an example, I would like to modify gdsc.settings.pvalue_correction_method from ‘fdr’ to ‘qvalue’, but when I set gdsc.settings.pvalue_correction_method=’qvalue’ and reissue the ANOVA(ic,gf), the method reverts to ‘fdr’. Also, drug names would be helpful, but I have yet to determine how to incorporate them when using the downloadable excel file Screened_Compounds.

3. When I attempt to run the python analysis in standalone the settings.json file does not appear to be read. I have placed the this file in the directory where I am running gdsctools, but, for example, when I specify the directory as ‘./standalone_html_gdsc_anova’ (or the complete pathname) I am getting no results in this directory. Perhaps this in linked to my difficulty when generating plots??

Regardless, I am committed to working this out, and appreciate your assistance.

Sincerely, David

From: Thomas Cokelaer [mailto:notifications@github.com] Sent: Thursday, March 02, 2017 4:42 AM To: CancerRxGene/gdsctools gdsctools@noreply.github.com Cc: Covell, David (NIH/NCI) [E] CovellD@mail.nih.gov; Author author@noreply.github.com Subject: Re: [CancerRxGene/gdsctools] settings (#166)

Hi David, I do not know off hand. In principle, the data should be in "/home/covelld/PYTHON/gdsctools-0.17.0/gdsctools/data/standalone_html_gdsc_anova" In which directory did you run the analysis ?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/CancerRxGene/gdsctools/issues/166#issuecomment-283605082, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AYVGfDbEFq_NIiRkOxQmpzwNot7TljbZks5rho7igaJpZM4MEnes.