Navigation

Clinical eslII (Essential Statistics Learning 2nd Edition)

In clinical studies, information about disease state and progression of a large number of patients is collected together with patient information that could (technically) be used for early diagnosis and or prognosis of disease. In this workflow clinical prostate cancer data of 97 patients is used to identify molecular markers that correlate with disease stage. The analysis is mainly inspired by the book ‘Essential Statistical Learning‘.

Analysis used in this workflow are cross-validation for Lasso penalized regression fit and best predictive variable identification (‘ncvreg’ and ‘leaps’ packages); linear regression and fitting with L1 constraints (‘stats’ and ‘lasso2’ packages); Lasso penalized Least Angle Regression with cross validation (‘lars’ package); and fitting General Linear Model (‘stats’ package).

digraph CLINICAL_eslII { {rank=same pros_regsubset pros_l1ce pros_lars pros_cvlars pros_glm} DataP [shape = invhouse, label = "data(prostate)"]; DataPTr [shape = invhouse, label = "Training set"]; DataPTe [shape = invhouse, label = "Test set"]; vs_ncvreg [shape = box; label = "Coordinate descent \ncv.ncvreg()"]; pros_regsubset [shape = box; label = "regsubset() "]; pros_l1ce [shape = box; label = "l1ce() "]; pros_lars [shape = box; label = "lars() "]; pros_cvlars [shape = box; label = "cv.lars() "]; pros_glm [shape = box; label = "glm() "]; pros_cvglm [shape = box; label = "cv.glm()"]; pros_rbind [label = "merge results \ndo.call(rbind)"]; pros_print [label = "print()"]; DataP -> DataPTr; DataP -> DataPTe; DataP -> vs_ncvreg -> pros_rbind -> pros_print; DataPTr -> pros_regsubset -> pros_rbind; DataPTr -> pros_l1ce -> pros_rbind; DataPTr -> pros_lars -> pros_rbind; DataPTr -> pros_cvlars -> pros_rbind; DataPTr -> pros_glm -> pros_rbind; pros_glm -> pros_cvglm -> pros_rbind; }

Workflow to identify prostate cancer markers using clinical data.

Packages and Dependencies

There are 6 packages used in this workflow, which depend on 2 additional packages from CRAN (dependencies)

Used packages:

  • CRAN: ncvreg, boot, lars, lasso2, mda, leaps

Package dependencies:

  • CRAN: class, MASS

Data

source datasets from http://cran.r-project.org/web/packages/lasso2/lasso2.pdf

  • Prostate dataset:
    • Hastie, T., Tibshirani, R., and Friedman, J. (2001). The Elements of Statistical Learning. Springer.
    • Stamey, T., et al. (1989). Prostate specific antigen in the diagnosis and treatment of adenocarcinoma of the prostate. II. Radical prostatectomy treated patients. Journal of Urology, 16: 1076-1083.

License

Table Of Contents

Previous topic

Affy

Next topic

Clinical (liver cohort)

This Page

Navigation

© Copyright 2016, BeDataDriven B.V.. Created using Sphinx 1.2.2.