Mercurial > repos > fubar > rglasso_1_9_8
diff rg_nri.xml @ 20:bb725f6d6d38 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/rglasso commit 344140b8df53b8b7024618bb04594607a045c03a
| author | iuc |
|---|---|
| date | Mon, 04 May 2015 22:47:29 -0400 |
| parents | 0e87f636bdd8 |
| children |
line wrap: on
line diff
--- a/rg_nri.xml Wed Apr 29 12:07:11 2015 -0400 +++ b/rg_nri.xml Mon May 04 22:47:29 2015 -0400 @@ -7,145 +7,12 @@ <requirement type="package" version="2.14">glmnet_lars_2_14</requirement> </requirements> <command interpreter="python"> - rgToolFactory.py --script_path "$runme" --interpreter "Rscript" --tool_name "rg_NRI" + rgToolFactory.py --script_path "$runme" --interpreter "Rscript" --tool_name "rg_NRI" --output_dir "$html_file.files_path" --output_html "$html_file" --make_HTML "yes" </command> - <inputs> - <param name="title" type="text" value="NRI test" size="80" label="Plot Title" help="Will appear as the title for the comparison plot"/> - <param name="input1" type="data" format="tabular" label="Select a tabular file from the baseline model with predicted and observed outcome column for each subject" - multiple='False' help="Observed and predicted status columns must be selected from this file below - NOTE both models must be in same order with exact matches in all observed outcomes" optional="False"/> - <param name="input1_observed" label="Select column containing observed outcome (0 for no event, 1 for an event)" type="data_column" data_ref="input1" numerical="True" - multiple="False" use_header_names="True" optional="False" help = "Observed outcomes are compared in the two files to check that the datasets are from the same data"/> - <param name="input1_predicted" label="Select column containing predicted event probabilies from baseline model" type="data_column" data_ref="input1" numerical="True" - multiple="False" use_header_names="True" optional="False" help="Must be in range 0-1"/> - <param name="input1_id" label="Select column containing subject ID from baseline model" type="data_column" data_ref="input1" numerical="True" - multiple="False" use_header_names="True" optional="False" help="Subect IDs are needed to match subjects to compare predictions in the two inputs"/> - <param name="input2" type="data" format="tabular" label="Select a tabular file from the new model with predicted and observed outcome columns for each subject" - multiple='False' help="Observed and predicted status columns must be selected from this file below" /> - <param name="input2_observed" label="Select column containing observed outcome (0 for no event, 1 for an event)" type="data_column" data_ref="input2" numerical="True" - multiple="False" use_header_names="True" optional="False" help = "Observed outcomes are compared in the two files to check that the datasets are from the same data"/> - <param name="input2_predicted" label="Select column containing predicted event probabilities from the new model" type="data_column" data_ref="input2" numerical="True" - multiple="False" use_header_names="True" optional="False" help="Must be in range 0-1"/> - <param name="input2_id" label="Select column containing subject ID from the new model" type="data_column" data_ref="input2" numerical="True" - multiple="False" use_header_names="True" optional="False" help="Subect IDs are needed to match subjects to compare predictions in the two inputs"/> - <conditional name="CImeth"> - <param name="cis" type="select" label="CI calculation method" - help="Bootstrap will take time - a long time for thousands - asymptotic is quick and informative"> - <option value="asymptotic" selected="true">Asymptotic estimate</option> - <option value="boot">Bootstrap for empirical CIs</option> - </param> - <when value="boot"> - <param name="nboot" type="integer" value="1000" label="Number of bootstrap replicates"/> - </when> - <when value="asymptotic"> - <param name="nboot" type="hidden" value="1000"/> - </when> - </conditional> - </inputs> - <outputs> - <data format="html" name="html_file" label="${title}.html"/> - <data format="tabular" name="nri_file" label="${title}_nrires.xls"/> - </outputs> - <tests> - <test> - <param name='title' value='nri_test1' /> - <param name='input1' value='nri_test1.xls' ftype='tabular' /> - <param name='input2' value='nri_test1.xls' ftype='tabular' /> - <param name='input1_id' value="1" /> - <param name='input1_observed' value="2" /> - <param name='input1_predicted' value="3" /> - <param name='input2_observed' value="2" /> - <param name='input2_predicted' value="4" /> - <output name='html_file' file='nri_test1_out.html' compare='diff' lines_diff='10' /> - <output name='nri_file' file='nri_test1_out.xls' /> - </test> -</tests> -<help> - -**Before you start** - -This is a simple tool to calculate various measures of improvement in prediction between two models described in pickering_paper_ -It is based on an R script pickering_code_ written by Dr John W Pickering and Dr David Cairns from sunny Otago University which -has been debugged and slightly adjusted to fit a Galaxy tool wrapper. - - -**What it does** - -Copied from the documentation in pickering_code_ :: - - - Functions to create risk assessment plots and associated summary statistics - - - (c) 2012 Dr John W Pickering, john.pickering@otago.ac.nz, and Dr David Cairns - Last modified August 2014 - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - * Redistributions of source code must retain the above copyright - notice, this list of conditions and the following disclaimer. - * Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in - the documentation and/or other materials provided with the distribution - - FUNCTIONS - raplot - Produces a Risk Assessment Plot and outputs the coordinates of the four curves - Based on: Pickering, J. W. and Endre, Z. H. (2012). New Metrics for Assessing Diagnostic Potential of - Candidate Biomarkers. Clinical Journal of the American Society of Nephrology, 7, 1355–1364. doi:10.2215/CJN.09590911 - - statistics.raplot - Produces the NRIs, IDIs, IS, IP, AUCs. - Based on: Pencina, M. J., D'Agostino, R. B. and Steyerberg, E. W. (2011). Extensions of net reclassification improvement calculations to - measure usefulness of new biomarkers. Statistics in Medicine, 30(1), 11–21. doi:10.1002/sim.4085 - Pencina, M. J., D'Agostino, R. B. and Vasan, R. S. (2008). Evaluating the added predictive ability of a new marker: From area under the - ROC curve to reclassification and beyond. - Statistics in Medicine, 27(2), 157–172. doi:10.1002/sim.2929 - DeLong, E., DeLong, D. and Clarke-Pearson, D. (1988). Comparing the areas under 2 or more correlated receiver operating characteristic curves - a nonparametric approach. - Biometrics, 44(3), 837–845. - - summary.raplot - Produces the NRIs, IDIs, IS, IP, AUCs with confidence intervals using a bootstrap or asymptotic procedure. (I prefer bootstrap which is chosed by cis=c("boot")) - - - Required arguments for all functions: - x1 is calculated risk (eg from a glm) for the null model, i.e. predict(,type="response") on a glm object - x2 is calculated risk (eg from a glm) for the alternative model - y is the case-control indicator (0 for controls, 1 for cases) - Optional argument - t are the boundaries of the risks for each group (ie 0, 1 and the thresholds beteween. eg c(0,0,3,0,7,1)). If missing, defaults to c(0, the incidence, 1) - - -**Input** - -The observed and predicted outcomes from two models to be compared. - -**Output** - -Lots'o'measures (TM) see pickering_paper_ for details - -**Attributions** - -pickering_paper_ is the paper the caclulations performed by this tool is based on - -pickering_code_ is the R function from John Pickering exposed by this Galaxy tool with minor modifications and hacks by Ross Lazarus. - -Galaxy_ (that's what you are using right now!) for gluing everything together - -Otherwise, all code and documentation comprising this tool was written by Ross Lazarus and is -licensed to you under the LGPL_ like other rgenetics artefacts - -.. _LGPL: http://www.gnu.org/copyleft/lesser.html -.. _pickering_code: http://www.researchgate.net/publication/264672640_R_function_for_Risk_Assessment_Plot__reclassification_metrics_NRI_IDI_cfNRI -.. _pickering_paper: http://cjasn.asnjournals.org/content/early/2012/05/24/CJN.09590911.full -.. _Galaxy: http://getgalaxy.org - - -</help> - <configfiles> <configfile name="runme"> - + <![CDATA[ ### http://www.researchgate.net/publication/264672640_R_function_for_Risk_Assessment_Plot__reclassification_metrics_NRI_IDI_cfNRI code @@ -159,13 +26,13 @@ ### ### (c) 2012 Dr John W Pickering, john.pickering@otago.ac.nz, and Dr David Cairns ### Last modified August 2014 -### -### Redistribution and use in source and binary forms, with or without +### +### Redistribution and use in source and binary forms, with or without ### modification, are permitted provided that the following conditions are met: -### * Redistributions of source code must retain the above copyright +### * Redistributions of source code must retain the above copyright ### notice, this list of conditions and the following disclaimer. -### * Redistributions in binary form must reproduce the above copyright -### notice, this list of conditions and the following disclaimer in +### * Redistributions in binary form must reproduce the above copyright +### notice, this list of conditions and the following disclaimer in ### the documentation and/or other materials provided with the distribution ### FUNCTIONS @@ -174,7 +41,7 @@ ### Based on: Pickering, J. W. and Endre, Z. H. (2012). New Metrics for Assessing Diagnostic Potential of ### Candidate Biomarkers. Clinical Journal of the American Society of Nephrology, 7, 1355–1364. doi:10.2215/CJN.09590911 ### -### statistics.raplot +### statistics.raplot ### Produces the NRIs, IDIs, IS, IP, AUCs. ### Based on: Pencina, M. J., D'Agostino, R. B. and Steyerberg, E. W. (2011). ### Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Statistics in Medicine, 30(1), 11–21. doi:10.1002/sim.4085 @@ -183,7 +50,7 @@ ### DeLong, E., DeLong, D. and Clarke-Pearson, D. (1988). Comparing the areas under 2 or more correlated receiver operating characteristic curves - a nonparametric approach. ### Biometrics, 44(3), 837–845. ### -### summary.raplot +### summary.raplot ### Produces the NRIs, IDIs, IS, IP, AUCs with confidence intervals using a bootstrap or asymptotic procedure. (I prefer bootstrap which is chosed by cis=c("boot")) @@ -229,13 +96,13 @@ ### actual plotting pdf(outplot) - plot(thresh.model1, sens.model1, xlim = c(0, 1), ylim = c(0, 1), type = "n", + plot(thresh.model1, sens.model1, xlim = c(0, 1), ylim = c(0, 1), type = "n", lty = 2, lwd = 2, xlab = "Risk of Event", ylab = "", col = "black", main=title) grid() - polygon(x = c(thresh.model1, thresh.model2[n.model2:1]), + polygon(x = c(thresh.model1, thresh.model2[n.model2:1]), y = c(sens.model1, sens.model2[n.model2:1]), border = NA, col = gray(0.8)) - polygon(x = c(thresh.model1, thresh.model2[n.model2:1]), + polygon(x = c(thresh.model1, thresh.model2[n.model2:1]), y = c(spec.model1, spec.model2[n.model2:1]), border = NA, col = gray(0.8)) lines(thresh.model1, sens.model1, type = "l", lty = 2, lwd = 2, col = "black") @@ -246,9 +113,9 @@ text(x = -0.15, y = 0.4, labels = "Sensitivity, ", col = "black", xpd = TRUE, srt = 90) text(x = -0.15, y = 0.4 + 0.175, labels = "1-Specificity", col = "red", xpd = TRUE, srt = 90) - legend("topleft", c("Event: New model", "Event: Baseline model", - "No Event: New model", "No Event: Baseline model"), - col = c("black", "black", "red", "red"), + legend("topleft", c("Event: New model", "Event: Baseline model", + "No Event: New model", "No Event: Baseline model"), + col = c("black", "black", "red", "red"), lty = c(1,2, 1, 2), lwd = 2, bty = "n") dev.off() return(data.frame("Null.p.sens"=thresh.model1, @@ -267,8 +134,8 @@ ### statistics from a raplot (is an adaptation of improveProb() from Hmisc) statistics.raplot = function(x1, x2, y, threshvec) -{ - +{ + s = is.na(x1 + x2 + y) ###Remove rows with missing data if (any(s)) { smiss = sum(s) @@ -302,7 +169,7 @@ na = sum(a) nb = sum(b) d = x2 - x1 - ### NRI + ### NRI n.thresh=length(threshvec)-1 risk.class.x1.ev=cut2(x1[a],threshvec) risk.class.x2.ev=cut2(x2[a],threshvec) @@ -317,18 +184,18 @@ pup.ev=0 pdown.ev=0 for (i in 1:(n.thresh-1)) { pup.ev = pup.ev + sum(cM.ev\$table[(i+1):n.thresh,i])} - for (i in 2:n.thresh) { pdown.ev = pdown.ev + sum(cM.ev\$table[1:(i-1),i])} - pup.ev=pup.ev/na + for (i in 2:n.thresh) { pdown.ev = pdown.ev + sum(cM.ev\$table[1:(i-1),i])} + pup.ev=pup.ev/na pdown.ev=pdown.ev/na risk.class.x1.ne=cut2(x1[b],threshvec) - risk.class.x2.ne=cut2(x2[b],threshvec) + risk.class.x2.ne=cut2(x2[b],threshvec) levels(risk.class.x1.ne)=thresh levels(risk.class.x2.ne)=thresh cM.ne=confusionMatrix(risk.class.x2.ne,risk.class.x1.ne) pup.ne=0 pdown.ne=0 for (i in 1:(n.thresh-1)){pup.ne=pup.ev+sum(cM.ne\$table[(i+1):n.thresh,i])} - for (i in 2:n.thresh){pdown.ne=pdown.ne+sum(cM.ne\$table[1:(i-1),i])} + for (i in 2:n.thresh){pdown.ne=pdown.ne+sum(cM.ne\$table[1:(i-1),i])} pdown.ne=pdown.ne/nb pup.ne=pup.ne/nb nri = pup.ev - pdown.ev - (pup.ne - pdown.ne) @@ -339,7 +206,7 @@ z.nri.ev = nri.ev/se.nri.ev nri.ne = pdown.ne - pup.ne se.nri.ne = sqrt((pdown.ne + pup.ne)/nb) - z.nri.ne = nri.ne/se.nri.ne + z.nri.ne = nri.ne/se.nri.ne ### Category Free NRI calculations cfpup.ev = mean(d[a] > 0) cfpup.ne = mean(d[b] > 0) @@ -400,31 +267,31 @@ ### Output output = c(n, na, nb, pup.ev, pup.ne, pdown.ev, pdown.ne, nri, se.nri, z.nri, - nri.ev, se.nri.ev, z.nri.ev, nri.ne, se.nri.ne, z.nri.ne, + nri.ev, se.nri.ev, z.nri.ev, nri.ne, se.nri.ne, z.nri.ne, cfpup.ev, cfpup.ne, cfpdown.ev, cfpdown.ne, cfnri, se.cfnri, z.cfnri, - cfnri.ev, se.cfnri.ev, z.cfnri.ev, cfnri.ne, se.cfnri.ne, z.cfnri.ne, - improveSens, improveSpec, idi.ev, se.idi.ev, z.idi.ev, idi.ne, - se.idi.ne, z.idi.ne, idi, se.idi, z.idi, is.x1, NA, is.x2, NA, - ip.x1, NA, ip.x2, NA, auc.x1, se.auc.x1, auc.x2, se.auc.x2, + cfnri.ev, se.cfnri.ev, z.cfnri.ev, cfnri.ne, se.cfnri.ne, z.cfnri.ne, + improveSens, improveSpec, idi.ev, se.idi.ev, z.idi.ev, idi.ne, + se.idi.ne, z.idi.ne, idi, se.idi, z.idi, is.x1, NA, is.x2, NA, + ip.x1, NA, ip.x2, NA, auc.x1, se.auc.x1, auc.x2, se.auc.x2, roc.test.x1.x2\$p.value,incidence) - names(output) = c("n", "na", "nb", "pup.ev", "pup.ne", "pdown.ev", "pdown.ne", + names(output) = c("n", "na", "nb", "pup.ev", "pup.ne", "pdown.ev", "pdown.ne", "nri", "se.nri", "z.nri", "nri.ev", "se.nri.ev", "z.nri.ev", "nri.ne", "se.nri.ne", "z.nri.ne", - "cfpup.ev", "cfpup.ne", "cfpdown.ev", "cfpdown.ne", + "cfpup.ev", "cfpup.ne", "cfpdown.ev", "cfpdown.ne", "cfnri", "se.cfnri", "z.cfnri", "cfnri.ev", "se.cfnri.ev", "z.cfnri.ev", "cfnri.ne", "se.cfnri.ne", "z.cfnri.ne", "improveSens", "improveSpec", - "idi.ev", "se.idi.ev", "z.idi.ev", "idi.ne", "se.idi.ne", - "z.idi.ne", "idi", "se.idi", "z.idi", "is.x1", "se.is.x1", - "is.x2", "se.is.x2", "ip.x1", "se.ip.x1", "ip.x2", "se.ip.x2", - "auc.x1", "se.auc.x1", "auc.x2", "se.auc.x2", + "idi.ev", "se.idi.ev", "z.idi.ev", "idi.ne", "se.idi.ne", + "z.idi.ne", "idi", "se.idi", "z.idi", "is.x1", "se.is.x1", + "is.x2", "se.is.x2", "ip.x1", "se.ip.x1", "ip.x2", "se.ip.x2", + "auc.x1", "se.auc.x1", "auc.x2", "se.auc.x2", "roc.test.x1.x2.pvalue","incidence") resdf = data.frame(N=n, Na=na, Nb=nb, pup.ev=pup.ev, pup.ne=pup.ne, pdown.ev=pdown.ev, pdown.ne=pdown.ne, NRI=nri, NRI.se=se.nri, NRI.z=z.nri, - NRI.ev=nri.ev, NRI.ev.se=se.nri.ev, NRI.ev.z=z.nri.ev, NRI.ne=nri.ne, NRI.ne.se=se.nri.ne, NRI.ne.z=z.nri.ne, + NRI.ev=nri.ev, NRI.ev.se=se.nri.ev, NRI.ev.z=z.nri.ev, NRI.ne=nri.ne, NRI.ne.se=se.nri.ne, NRI.ne.z=z.nri.ne, cfpup.ev=cfpup.ev, cfpup.ne=cfpup.ne, cfpdown.ev=cfpdown.ev, cfpdown.ne=cfpdown.ne, CFNRI=cfnri, CFNRI.se=se.cfnri, CFNRI.z=z.cfnri, - CFNRI.ev=cfnri.ev, CFNRI.ev.se=se.cfnri.ev, CFNRI.ev.z=z.cfnri.ev, CFNRI.ne=cfnri.ne, CFNRI.ne.se=se.cfnri.ne, CFNRI.ne.z=z.cfnri.ne, - improvSens=improveSens, improvSpec=improveSpec, IDI.ev=idi.ev, IDI.ev.se=se.idi.ev, IDI.ev.z=z.idi.ev, IDI.ne=idi.ne, - IDI.ne.se=se.idi.ne, IDI.ne.z=z.idi.ne, IDI=idi, IDI.se=se.idi, IDI.z=z.idi, isx1=is.x1, isx2=is.x2, - ipxi=ip.x1, ipx2=ip.x2, AUC.x1=auc.x1, AUC.x1.se=se.auc.x1, AUC.x2=auc.x2, AUC.x2.se=se.auc.x2, + CFNRI.ev=cfnri.ev, CFNRI.ev.se=se.cfnri.ev, CFNRI.ev.z=z.cfnri.ev, CFNRI.ne=cfnri.ne, CFNRI.ne.se=se.cfnri.ne, CFNRI.ne.z=z.cfnri.ne, + improvSens=improveSens, improvSpec=improveSpec, IDI.ev=idi.ev, IDI.ev.se=se.idi.ev, IDI.ev.z=z.idi.ev, IDI.ne=idi.ne, + IDI.ne.se=se.idi.ne, IDI.ne.z=z.idi.ne, IDI=idi, IDI.se=se.idi, IDI.z=z.idi, isx1=is.x1, isx2=is.x2, + ipxi=ip.x1, ipx2=ip.x2, AUC.x1=auc.x1, AUC.x1.se=se.auc.x1, AUC.x2=auc.x2, AUC.x2.se=se.auc.x2, roctestpval=roc.test.x1.x2\$p.value,incidence=incidence) tr = t(resdf) tresdf = data.frame(measure=colnames(resdf),value=tr[,1]) @@ -453,7 +320,7 @@ boot.index = sample(length(y), replace = TRUE) risk.model1.boot = x1[boot.index] risk.model2.boot = x2[boot.index] - cc.status.boot = y[boot.index] + cc.status.boot = y[boot.index] r = statistics.raplot(x1 = risk.model1.boot, x2 = risk.model2.boot, y = cc.status.boot) results.boot[i, ] = r\$output } @@ -478,87 +345,87 @@ results.matrix[2, ] = c("Events (n)", results["na"]) results.matrix[3, ] = c("Non-events (n)", results["nb"]) results.matrix[4, ] = c("Category free NRI and summary statistics","-------------------------") - results.matrix[5, ] = c("cfNRI events (%)", - paste(round(100*results["cfnri.ev"], dp-2), " (", + results.matrix[5, ] = c("cfNRI events (%)", + paste(round(100*results["cfnri.ev"], dp-2), " (", round(100*results["cfnri.ev"] - z * 100*results["se.cfnri.ev"], dp-2), - " to ", round(100*results["cfnri.ev"] + + " to ", round(100*results["cfnri.ev"] + z * 100*results["se.cfnri.ev"], dp-2), ")", sep = "")) - results.matrix[6, ] = c("cfNRI non-events (%)", + results.matrix[6, ] = c("cfNRI non-events (%)", paste(round(100*results["cfnri.ne"], dp-2), " (", round(100*results["cfnri.ne"] - z * 100*results["se.cfnri.ne"], dp)-2, - " to ", round(100*results["cfnri.ne"] + z * 100*results["se.cfnri.ne"], - dp-2), ")", sep = "")) - results.matrix[7, ] = c("cfNRI (%)", - paste(round(100*results["cfnri"], dp-2), " (", - round(100*results["cfnri"] - z * 100*results["se.cfnri"], dp-2), - " to ", round(100*results["cfnri"] + z * 100*results["se.cfnri"], + " to ", round(100*results["cfnri.ne"] + z * 100*results["se.cfnri.ne"], + dp-2), ")", sep = "")) + results.matrix[7, ] = c("cfNRI (%)", + paste(round(100*results["cfnri"], dp-2), " (", + round(100*results["cfnri"] - z * 100*results["se.cfnri"], dp-2), + " to ", round(100*results["cfnri"] + z * 100*results["se.cfnri"], dp-2), ")", sep = "")) results.matrix[8, ] = c("NRI and summary statistics","-------------------------") - results.matrix[9, ] = c("NRI events (%)", - paste(round(100*results["nri.ev"], dp-2), " (", + results.matrix[9, ] = c("NRI events (%)", + paste(round(100*results["nri.ev"], dp-2), " (", round(100*results["nri.ev"] - z * 100*results["se.nri.ev"], dp-2), - " to ", round(100*results["nri.ev"] + + " to ", round(100*results["nri.ev"] + z * 100*results["se.nri.ev"], dp-2), ")", sep = "")) - results.matrix[10, ] = c("NRI non-events (%)", + results.matrix[10, ] = c("NRI non-events (%)", paste(round(100*results["nri.ne"], dp-2), " (", round(100*results["nri.ne"] - z * 100*results["se.nri.ne"], dp-2), - " to ", round(100*results["nri.ne"] + z * 100*results["se.nri.ne"], - dp-2), ")", sep = "")) - results.matrix[11, ] = c("NRI (%)", - paste(round(100*results["nri"], dp-2), " (", - round(100*results["nri"] - z * 100*results["se.nri"], dp-2), - " to ", round(100*results["nri"] + z * 100*results["se.nri"], + " to ", round(100*results["nri.ne"] + z * 100*results["se.nri.ne"], + dp-2), ")", sep = "")) + results.matrix[11, ] = c("NRI (%)", + paste(round(100*results["nri"], dp-2), " (", + round(100*results["nri"] - z * 100*results["se.nri"], dp-2), + " to ", round(100*results["nri"] + z * 100*results["se.nri"], dp-2), ")", sep = "")) results.matrix[12, ] = c("IDI and summary statistics","-------------------------") - results.matrix[13, ] = c("IDI events", - paste(round(results["idi.ev"], dp), " (", - round(results["idi.ev"] - z * results["se.idi.ev"], dp), - " to ", round(results["idi.ev"] + z * results["se.idi.ev"], + results.matrix[13, ] = c("IDI events", + paste(round(results["idi.ev"], dp), " (", + round(results["idi.ev"] - z * results["se.idi.ev"], dp), + " to ", round(results["idi.ev"] + z * results["se.idi.ev"], dp), ")", sep = "")) - results.matrix[14, ] = c("IDI non-events", - paste(round(results["idi.ne"], dp), " (", - round(results["idi.ne"] - z * results["se.idi.ne"], dp), - " to ", round(results["idi.ne"] + z * results["se.idi.ne"], + results.matrix[14, ] = c("IDI non-events", + paste(round(results["idi.ne"], dp), " (", + round(results["idi.ne"] - z * results["se.idi.ne"], dp), + " to ", round(results["idi.ne"] + z * results["se.idi.ne"], dp), ")", sep = "")) - results.matrix[15, ] = c("IDI", - paste(round(results["idi"], dp), " (", - round(results["idi"] - z * results["se.idi"], dp), - " to ", round(results["idi"] + z * results["se.idi"], + results.matrix[15, ] = c("IDI", + paste(round(results["idi"], dp), " (", + round(results["idi"] - z * results["se.idi"], dp), + " to ", round(results["idi"] + z * results["se.idi"], dp), ")", sep = "")) - results.matrix[16, ] = c("IS (null model)", - paste(round(results["is.x1"], dp), " (", - round(results["is.x1"] - z * results["se.is.x1"], dp), - " to ", round(results["is.x1"] + z * results["se.is.x1"], + results.matrix[16, ] = c("IS (null model)", + paste(round(results["is.x1"], dp), " (", + round(results["is.x1"] - z * results["se.is.x1"], dp), + " to ", round(results["is.x1"] + z * results["se.is.x1"], dp), ")", sep = "")) - results.matrix[17, ] = c("IS (alt model)", - paste(round(results["is.x2"], dp), " (", - round(results["is.x2"] - z * results["se.is.x2"], dp), - " to ", round(results["is.x2"] + z * results["se.is.x2"], + results.matrix[17, ] = c("IS (alt model)", + paste(round(results["is.x2"], dp), " (", + round(results["is.x2"] - z * results["se.is.x2"], dp), + " to ", round(results["is.x2"] + z * results["se.is.x2"], dp), ")", sep = "")) - results.matrix[18, ] = c("IP (null model)", - paste(round(results["ip.x1"], dp), " (", - round(results["ip.x1"] - z * results["se.ip.x1"], dp), - " to ", round(results["ip.x1"] + z * results["se.ip.x1"], + results.matrix[18, ] = c("IP (null model)", + paste(round(results["ip.x1"], dp), " (", + round(results["ip.x1"] - z * results["se.ip.x1"], dp), + " to ", round(results["ip.x1"] + z * results["se.ip.x1"], dp), ")", sep = "")) - results.matrix[19, ] = c("IP (alt model)", - paste(round(results["ip.x2"], dp), " (", - round(results["ip.x2"] - z * results["se.ip.x2"], dp), - " to ", round(results["ip.x2"] + z * results["se.ip.x2"], + results.matrix[19, ] = c("IP (alt model)", + paste(round(results["ip.x2"], dp), " (", + round(results["ip.x2"] - z * results["se.ip.x2"], dp), + " to ", round(results["ip.x2"] + z * results["se.ip.x2"], dp), ")", sep = "")) results.matrix[20, ] = c("AUC","-------------------------") - results.matrix[21, ] = c("AUC (null model)", - paste(round(results["auc.x1"], dp), " (", - round(results["auc.x1"] - z * results["se.auc.x1"], dp), - " to ", round(results["auc.x1"] + z * results["se.auc.x1"], + results.matrix[21, ] = c("AUC (null model)", + paste(round(results["auc.x1"], dp), " (", + round(results["auc.x1"] - z * results["se.auc.x1"], dp), + " to ", round(results["auc.x1"] + z * results["se.auc.x1"], dp), ")", sep = "")) - results.matrix[22, ] = c("AUC (alt model)", - paste(round(results["auc.x2"], dp), " (", - round(results["auc.x2"] - z * results["se.auc.x2"], dp), - " to ", round(results["auc.x2"] + z * results["se.auc.x2"], + results.matrix[22, ] = c("AUC (alt model)", + paste(round(results["auc.x2"], dp), " (", + round(results["auc.x2"] - z * results["se.auc.x2"], dp), + " to ", round(results["auc.x2"] + z * results["se.auc.x2"], dp), ")", sep = "")) results.matrix[23, ] = c("difference (P)", round(results["roc.test.x1.x2.pvalue"], dp)) results.matrix[24, ] = c("Incidence", round(results["incidence"], dp)) - + return(results.matrix) } @@ -624,6 +491,139 @@ sink() </configfile> </configfiles> + <inputs> + <param name="title" type="text" value="NRI test" size="80" label="Plot Title" help="Will appear as the title for the comparison plot"/> + <param name="input1" type="data" format="tabular" label="Select a tabular file from the baseline model with predicted and observed outcome column for each subject" + multiple='False' help="Observed and predicted status columns must be selected from this file below - NOTE both models must be in same order with exact matches in all observed outcomes" optional="False"/> + <param name="input1_observed" label="Select column containing observed outcome (0 for no event, 1 for an event)" type="data_column" data_ref="input1" numerical="True" + multiple="False" use_header_names="True" optional="False" help = "Observed outcomes are compared in the two files to check that the datasets are from the same data"/> + <param name="input1_predicted" label="Select column containing predicted event probabilies from baseline model" type="data_column" data_ref="input1" numerical="True" + multiple="False" use_header_names="True" optional="False" help="Must be in range 0-1"/> + <param name="input1_id" label="Select column containing subject ID from baseline model" type="data_column" data_ref="input1" numerical="True" + multiple="False" use_header_names="True" optional="False" help="Subect IDs are needed to match subjects to compare predictions in the two inputs"/> + <param name="input2" type="data" format="tabular" label="Select a tabular file from the new model with predicted and observed outcome columns for each subject" + multiple='False' help="Observed and predicted status columns must be selected from this file below" /> + <param name="input2_observed" label="Select column containing observed outcome (0 for no event, 1 for an event)" type="data_column" data_ref="input2" numerical="True" + multiple="False" use_header_names="True" optional="False" help = "Observed outcomes are compared in the two files to check that the datasets are from the same data"/> + <param name="input2_predicted" label="Select column containing predicted event probabilities from the new model" type="data_column" data_ref="input2" numerical="True" + multiple="False" use_header_names="True" optional="False" help="Must be in range 0-1"/> + <param name="input2_id" label="Select column containing subject ID from the new model" type="data_column" data_ref="input2" numerical="True" + multiple="False" use_header_names="True" optional="False" help="Subect IDs are needed to match subjects to compare predictions in the two inputs"/> + <conditional name="CImeth"> + <param name="cis" type="select" label="CI calculation method" + help="Bootstrap will take time - a long time for thousands - asymptotic is quick and informative"> + <option value="asymptotic" selected="true">Asymptotic estimate</option> + <option value="boot">Bootstrap for empirical CIs</option> + </param> + <when value="boot"> + <param name="nboot" type="integer" value="1000" label="Number of bootstrap replicates"/> + </when> + <when value="asymptotic"> + <param name="nboot" type="hidden" value="1000"/> + </when> + </conditional> + </inputs> + <outputs> + <data format="html" name="html_file" label="${title}.html"/> + <data format="tabular" name="nri_file" label="${title}_nrires.xls"/> + </outputs> + <tests> + <test> + <param name='title' value='nri_test1' /> + <param name='input1' value='nri_test1.xls' ftype='tabular' /> + <param name='input2' value='nri_test1.xls' ftype='tabular' /> + <param name='input1_id' value="1" /> + <param name='input1_observed' value="2" /> + <param name='input1_predicted' value="3" /> + <param name='input2_observed' value="2" /> + <param name='input2_predicted' value="4" /> + <output name='html_file' file='nri_test1_out.html' compare='diff' lines_diff='10' /> + <output name='nri_file' file='nri_test1_out.xls' /> + </test> +</tests> +<help> + +**Before you start** + +This is a simple tool to calculate various measures of improvement in prediction between two models described in pickering_paper_ +It is based on an R script pickering_code_ written by Dr John W Pickering and Dr David Cairns from sunny Otago University which +has been debugged and slightly adjusted to fit a Galaxy tool wrapper. + + +**What it does** + +Copied from the documentation in pickering_code_ :: + + + Functions to create risk assessment plots and associated summary statistics + + + (c) 2012 Dr John W Pickering, john.pickering@otago.ac.nz, and Dr David Cairns + Last modified August 2014 + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in + the documentation and/or other materials provided with the distribution + + FUNCTIONS + raplot + Produces a Risk Assessment Plot and outputs the coordinates of the four curves + Based on: Pickering, J. W. and Endre, Z. H. (2012). New Metrics for Assessing Diagnostic Potential of + Candidate Biomarkers. Clinical Journal of the American Society of Nephrology, 7, 1355–1364. doi:10.2215/CJN.09590911 + + statistics.raplot + Produces the NRIs, IDIs, IS, IP, AUCs. + Based on: Pencina, M. J., D'Agostino, R. B. and Steyerberg, E. W. (2011). Extensions of net reclassification improvement calculations to + measure usefulness of new biomarkers. Statistics in Medicine, 30(1), 11–21. doi:10.1002/sim.4085 + Pencina, M. J., D'Agostino, R. B. and Vasan, R. S. (2008). Evaluating the added predictive ability of a new marker: From area under the + ROC curve to reclassification and beyond. + Statistics in Medicine, 27(2), 157–172. doi:10.1002/sim.2929 + DeLong, E., DeLong, D. and Clarke-Pearson, D. (1988). Comparing the areas under 2 or more correlated receiver operating characteristic curves - a nonparametric approach. + Biometrics, 44(3), 837–845. + + summary.raplot + Produces the NRIs, IDIs, IS, IP, AUCs with confidence intervals using a bootstrap or asymptotic procedure. (I prefer bootstrap which is chosed by cis=c("boot")) + + + Required arguments for all functions: + x1 is calculated risk (eg from a glm) for the null model, i.e. predict(,type="response") on a glm object + x2 is calculated risk (eg from a glm) for the alternative model + y is the case-control indicator (0 for controls, 1 for cases) + Optional argument + t are the boundaries of the risks for each group (ie 0, 1 and the thresholds beteween. eg c(0,0,3,0,7,1)). If missing, defaults to c(0, the incidence, 1) + + +**Input** + +The observed and predicted outcomes from two models to be compared. + +**Output** + +Lots'o'measures (TM) see pickering_paper_ for details + +**Attributions** + +pickering_paper_ is the paper the caclulations performed by this tool is based on + +pickering_code_ is the R function from John Pickering exposed by this Galaxy tool with minor modifications and hacks by Ross Lazarus. + +Galaxy_ (that's what you are using right now!) for gluing everything together + +Otherwise, all code and documentation comprising this tool was written by Ross Lazarus and is +licensed to you under the LGPL_ like other rgenetics artefacts + +.. _LGPL: http://www.gnu.org/copyleft/lesser.html +.. _pickering_code: http://www.researchgate.net/publication/264672640_R_function_for_Risk_Assessment_Plot__reclassification_metrics_NRI_IDI_cfNRI +.. _pickering_paper: http://cjasn.asnjournals.org/content/early/2012/05/24/CJN.09590911.full +.. _Galaxy: http://getgalaxy.org + + +</help> + <citations> <citation type="doi">doi: 10.2215/CJN.09590911</citation> </citations>