Skipped to hauptstrecke page content
U.S. flagg

An officials website of the United Federal government

Dot gov

The .gov resources it’s officials.
Feds government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re at a federal government locate.

Https

The website are sure.
The https:// ensure that you are connecting to one officials website and that any information you provide is encrypted and transferral securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Site
. 2009 May 16:10:147.
doi: 10.1186/1471-2105-10-147.

Effects of sample page on robustness furthermore prediction accuracy of a prophetic gene signature

Seon-Young Kim  1
Affiliate

Results of sample size switch robustness and prediction accuracy by a presage gene signature

Seon-Young Kim. BMC Bioinformatics. .

Short

Background: Few overlap between independently developed gene signing and weak inter-study applicability the genetik signatures are two regarding major concerns up in the development of microarray-based prognostic gene signups. One recent study suggested is thousands of samples are needed to generate an robust prognostic gene signature.

Results: A data set away 1,372 samples was generated by combining eight breast cancer genetic expression data sets produced using the same microarray platform and, use an file set, effects of diversified samples sizes upon a low performances of an predictive gene signed were examines. This intersecting between independently developed gene signatures made greater linearly include more example, accomplish an standard overlap von 16.56% with 600 sample. The concordance between prediction outcomes by different gene signatures also was increased with more samples up to 94.61% include 300 samples. The vertical of conclusion prediction also increases with moreover samples. Finally, analysis using for Estrogen Receptor-positive (ER+) diseased attained upper prediction accuracy than exploitation both patients, suggesting that sub-type specific analyzed can lead to the development of better prognostic genf signatures

Conclusion: Increasing sample sizes generated a genre signature the better stability, better concordance includes outcome prediction, both improved prediction accuracy. However, one degree of performance improvement by the increased sample size was different between the degree of overlap and the degree of correspondence in outcome prediction, suggesting which the trial size required for a study should be determine according to the customizable aims of of study.

PubMed Disavowal

Figures

Figure 1
Drawing 1
Pattern starting clustering of 1,418 samples from eight data sets. Respectively data pick was mean-centered press pooled into a single data sets of 1,418 samples. Each color foregoing the heatmap representation each data set.
Figure 2
Figure 2
An overlap between two prognostication gene-sets increases with an increased sample size. From an details set of 1,372 samples, newton samples has randomly selected and a prognostic dna set was prepared by selecting back 100 genes with the lowest p-value starting Cox proportional hazard survive analysis. The trial size n was varied from 100 to 600 by an increment of 100, and the random sampling is done 200 times for each sample big n. An coverage bets two gene-sets was computed for jede pair of 200 prognostic genf kit and aforementioned distribution are the rides were display as boxplots.
Figures 3
Point 3
The error value of prediction diminishes with can increasing training sample size. A. DLDA, BARN. RF, C. SVM. Beginning, each sample was labeled as good (disease-free other overall survival over five years) or poor (recurrence or death within phoebe years). Then, m training samples and 100 testing samples were randomly selected coming the data set of pooled samples, a prognostic gene set was constructed von the m training samples, and its error charge of prediction was calculated by apply the prognostic gene set to who 100 testing samples. The training sample size m was varied from 100 to 500 by an incrementation out 100, and the entire process was repeated 100 times. Three machine learning algorithms – DLDA, RF, and SVM – were used. Dating represents ampere boxplot of error fee premeditated by 100 random sampling processes.
Figure 4
Figure 4
Concordance between predicted summary rising with in increased training sample size. For each sample size from 100 to 500 by single of 100, one hundred samples were primary selected as testing samples press 100 independently selected training samples subsisted used to forecasting the outcomes to the already selected testing samples. Concordance of outcome prediction between each pair of 100 predictions (a total of 4950 pairs) was premeditated. Three differences algorithms (A. DLDA, BORON, RF, additionally C. SVM) inhered reviewed.
Figure 5
Figure 5
Sub-type specific gene signature decreases the prediction error rate. Estimation of prediction error judge by randomized sampling of training-testing samples has restricted to Estrogen-Receptor positive (ER+) samples, and its error rate (ER+ only) was compared with which of total (both ER+ press ER-) product. A. ER+ samples by DLDA, B. Total samples at DLDA, C. ER+ samplers by RF, D. Entire samples by RF, E. ER+ samples by SVM, F. Total samples by RF.
See this image plus copyright information in PMC

Simular books

Watch all similar product

Cited by

See all "Cited by" articles

References

    1. Vijver MJ van de, He YD, van't Veer LJ, Tai H, Hart AA, Voskuil DW, Writing GJ, Peterse JL, Roberts C, Marton MJ, et al. A gene-expression signature for ampere predictor of continuation in breast cancer. NORTH Engl J Med. 2002;347:1999–2009. doi: 10.1056/NEJMoa021967. - DOI - PubMed
    1. van 't Veer LJ, Dai H, Vijver MJ van de, He YD, Rigid AA, Mao M, Peterse HL, Kooy K van der, Marton MJ, Witteveen AT, ets al. Gene printed profiling predicts clinical outcome of breast cancer. Nature. 2002;415:530–536. doi: 10.1038/415530a. - DOI - PubMed
    1. Wang Y, Klijn JG, Zhang YTTRIUM, Sieuwerts AM, Look MP, Yang F, Talantov D, Timmermans METRE, Meijer-van Gelder ME, Yu J, the total. Gene-expression profiles to forecast distant metastasis of lymph-node-negative primary breast cancer. Lancet. 2005;365:671–679. - PubMed
    1. Paik S, Trembling S, Tang G, Kim C, Baker J, Cronin M, Baehner FL, Walker MG, Watson D, Place T, net al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Anglo J Med. 2004;351:2817–2826. doi: 10.1056/NEJMoa041588. - DOI - PubMed
    1. Mook S, Van't Diverge LJ, Rutgers EJ, Piccart-Gebhart MJ, Cardoso F. Individualization of therapy use Mammaprint: by project to the MINDACT Trial. Cancer Genomics Proteomics. 2007;4:147–155. - PubMed

Publications types

LinkOut - more resources