return to case studies

Analysis of Golub and Slonim Data

Client

AnVil examined the dataset from Golub and Slonim, et al. (1999), where they collected a group of 72 tissue samples taken from patients with 2 subtypes of acute leukemia, and conducted DNA microarray analysis of gene expression profiles. This data set has become a common starting point for comparison of microarray analytical methods.

Read the original paper (PDF format) by Golub & Slonim, et al. for more details.

Challenge

AnVil undertook this study to demonstrate how high-dimensional visualization of massive microarray data sets can reveal valuable patterns and relationships in complete, unreduced data sets, to guide subsequent analysis efforts. Golub & Slonim, et al. produced gene expression profiles from 72 acute leukemia patient samples gathered from four sites over a period of almost 20 years, using Affymetrix microarrays and data preparation software. While the lack of replicate testing and sample quality control may have increased the variability in the resulting data set, the data were sufficient for the authors to successfully classify cancer types and discover cancer sub-types.

Objectives

AnVil’s goal was to use its novel approach to find answers in the data where other groups have fallen short. AnVil's approach was set apart from others by the use of the visual methods we used to conduct our analyses and our big picture view of the problem, as well as the rigor with which we verified our answers.

Solutions

AnVil's approach preserved the value and meaning inherent in the full data set by creating global data and meta-statistical overviews that enabled us to reveal major data patterns and identify aberrant samples that may bias results. Departing from the now traditional analysis methods exemplified by Golub & Slonim, et al., AnVil first took a high-level overview of the data set, examining statistical metadata and working in high-dimensional space to assess the full data set. Visualizations were used not only as a means of portraying the results of analyses, but also as interactive tools for the exploration, manipulation and analysis of the data and generation of subsequent results.

Results

AnVil prepared a 76-gene classifier that was shown to be completely accurate in classifying the 72 leukemia samples using several machine learning methods. (See Figure 1) Another 25 leukemia samples from a completely independent study (Virtenave, et al. (2000) were classified with 90 percent accuracy using this 76-gene predictor set. Analysis of limited chemotherapy treatment outcome data also yielded a 76-gene predictor for remission success and failure; a finding until recently unique among the dozens of presented analyses of the Golub & Slonim, et al. data set. (See Figure 2) This work demonstrates the applicability of AnVil’s technology and processes in efforts to construct DNA microarray “disease chips” for use in diagnostics.

Figure 1 Click to Enlarge	Figure 2 Click to Enlarge

Read more details about AnVil's analysis of the Golub & Slonim data set.

back to top