This article is part of the supplement: Italian Society of Bioinformatics (BITS): Annual Meeting 2011

Open Access Research

Design of a multi-signature ensemble classifier predicting neuroblastoma patients' outcome

Andrea Cornero1, Massimo Acquaviva1, Paolo Fardin1, Rogier Versteeg2, Alexander Schramm3, Alessandra Eva1, Maria Carla Bosco1, Fabiola Blengio1, Sara Barzaghi1 and Luigi Varesio1*

Author Affiliations

1 Laboratory of Molecular Biology, G. Gaslini Institute, Genoa 16147, Italy

2 Department of Human Genetics, Academic Medical Center, University of Amsterdam, Amsterdam 1100, The Netherlands

3 Department of Pediatric Oncology and Hematology, University Children's Hospital Essen, Essen 45122, Germany

For all author emails, please log on.

BMC Bioinformatics 2012, 13(Suppl 4):S13  doi:10.1186/1471-2105-13-S4-S13

Published: 28 March 2012

Additional files

Additional file 1:

External validation of single signature classifiers. Predictions performed on DS2, (60 patients) dataset by the 20 signatures associated to the best individual classifiers selected during the first phase are assembled in a prediction matrix and shown together with the true outcome. A: alive, D: dead. The signatures are indicated by the initials of the first author of the manuscript. The associated references are as follows: DIP: Di Pietro [23], FI: Fischer [8], FR: Fredlund [25], NE I: Nevo [30], OB: Oberhuer [7], OH: Ohira [6], WE: Wei [4], DEP I: De Preter [10], FA: Fardin [12], OE: Oe [28], SH: Shimada [27], HA: Hahn [26], CH: Chen [22], DEP II: De Preter II [1], BE: Benard [21], FR: Fransson [24], MCA: McArdle [29], AS: Asgharzadeh [20], NE II: Nevo II [31], VE: Vermulen [9].

Format: XLS Size: 39KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 2:

NB-MuSE-classifier external validation. Predictions performed on DS3, 62 patients dataset. The best individual-signature classifiers selected during the first phase have been assembled in a prediction matrix used for the external validation of the NB-MuSE-classifier. The "Actual" column represent the true Alive/Dead label for each patient. The "MuSE" column represent the outcome of the NB-MuSE-classifier in external validation. A: alive, D: dead. The signatures are indicated by the initials of the first author of the manuscript (see legend to additional file 1).

Format: XLS Size: 41KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 3:

Kaplan-Meier and log-rank analysis of patients with Stage 4 tumors stratified according to the NB-MuSE classifier. Kaplan-Meier and log-rank analysis for INSS Stage 4 neuroblastoma patients belonging to the external validation dataset. 5-years overall survival (left) and event free survival (right) of patients stratified according to the NB-MuSE classifier. Red and black curves represent poor and good outcome patients respectively. The p-value of the log-rank test is shown.

Format: PDF Size: 93KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 4:

Kaplan-Meier and log-rank analysis of patients with localized and Stage 4s tumors stratified according to the NB-MuSE classifier. Kaplan-Meier and log-rank analysis for INSS Stage 1,2,3 and 4s neuroblastoma patients belonging to the external validation dataset. 5-years overall survival (left) and event free survival (right) of patients stratified according to the NB-MuSE classifier. Red and black curves represent poor and good outcome patients respectively. The p-value of the log-rank test is shown.

Format: PDF Size: 77KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 5:

Kaplan-Meier and log-rank analysis of patients with MYCN amplified tumors stratified according to the NB-MuSE classifier. Kaplan-Meier and log-rank analysis for neuroblastoma patients with MYCN amplified tumors belonging to the external validation dataset. 5-years overall survival (left) and event free survival (right) of patients stratified according to the NB-MuSE classifier. Red and black curves represent poor and good outcome patients respectively. The p-value of the log-rank test is shown.

Format: PDF Size: 79KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 6:

Kaplan-Meier and log-rank analysis of patients without MYCN amplification in the tumors stratified according to the NB-MuSE classifier. Kaplan-Meier and log-rank analysis for neuroblastoma patients with MYCN not amplified tumors belonging to the external validation dataset. 5-years overall survival (left) and event free survival (right) of patients stratified according to the NB-MuSE classifier. Red and black curves represent poor and good outcome patients respectively. The p-value of the log-rank test is shown.

Format: PDF Size: 76KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 7:

Probesets and genes composition of the 20 signatures selected for building the MuSE classifier. 20 signatures, out of 33 tested, were selected representing those performing with an accuracy > = 80. The gene composition (Gene Name) and matching probeset (Affymetrix) are shown. The choice of the probesets was made according to the criteria described in the Material and Methods' section. The signatures are indicated by the initials of the first author of the manuscript (see legend to additional file 1).

Format: XLS Size: 66KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 8:

Best performing algorithms tested to build the MuSE classifier. The classification accuracy of the models that were used in the process leading to the selection of the best performing algorithm are shown together with the parameters describing the performances of each. The 18 algorithms that gave a meaningful separation are reported. ClassificationViaClustering, FT, BayesLogisticRegression and MBtree did not give a significant separation and are not shown. The algorithms are described and implemented in WEKA package [44]. The signatures are indicated by the initials of the first author of the manuscript (see legend to additional file 1).

Format: XLS Size: 30KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 9:

Characteristics of the datasets. The 182 patients dataset was divided randomly into three subsets (DS1, DS2, and DS3) consisting of 60, 60, and 62 patients respectively. The characteristics of the composition of these datasets are shown. For each DS, the composition in terms of INSS stage representation, MYCN amplification, source of the data (AMC: Amsterdam; ESS: Essen; HI: Hiroshima; IGG: Genoa, for details see Material and Methods), age at diagnosis and Overall survival.

Format: XLS Size: 29KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data