...
Eric painstakingly generated paired data for about 1,400 spots from the segmented images and their paired spot lists for a few gels. Because of the difficulty in matching or pairing spots between gels, I also developed spot-pairing and spot "landmarking" programs, which ran interactively on the RTPP/PDP8e. (Landmarking is visually identifying a set of spots that are common in both the reference gel and each additional gel sample so spot pairing could proceed.) Eric then subjected the data to SPSS statistical analysis with encouraging results (t-Tests, Spearman correlation, ANOVA) and it was published in [14] and enhancements in [15,18-21]. All three of these programs were rewritten in SAIL for the DECsystem-2020. At this point we realized that we wanted to build a database containing large numbers of gels to detect marker proteins or classify samples by protein pattern signatures indicating states of differentiation, disease conditions, or other experimental conditions. This was the basis of GELLAB-I.
This led to the SAIL program CGELP, in GELLAB-I, to construct a composite gel database with a virtual reference gel [14-18,20,30] and the addition of many more statistical methods. (The subsequent Unix version was called CGELP2 [30, 42, 43, 44, 45] where additional statistical exploratory data analysis methods were added - see [history of GELLAB ].) Spots of a set of N-1 gels would be matched to one of the gels called a reference gel and spots missing in the physical reference gel would be extrapolated into the reference gel. Figure 19 shows the reference gel 324.1 that was an acute myeloid leukemia (AML) gel, scanned with the RTPP. This reference gel was used in many of the leukemia databases to tie the data together [14, 15, 16, 17, 18, 22]. The collection of SAIL programs, as well as their RTPP interface, was called GELLAB-I. After leaving NIH (for the University of Chicago), Eric would fly back to work in our laboratory to do marathon late-night landmarking sessions to help generate these databases of large numbers of 2D leukemia gels. Over the years, we had built various databases with over 400 gel samples. The leukemia database had over 130 samples. Some of these gel sample images are available on the bioinformatics.org/lecb2dgeldb open source repository. This early research led to my interest in exploratory data analysis and future work with microarrays with MAExplorer.sourceforge.net [46], and proteomics exploratory data analysis using open2dprot.sourceforge.net .
Peter Sonderegger, while a post-doc at the National Institute of Child Health and Development (NICHD), used the GELLAB-I system with the RTPP to investigate how the expression of axonal proteins of sensory and motor neurons was influenced by non-neuronal cells [27, 29, 45]. At the same time we were investigating the feasibility of porting GELLAB-I (written in SAIL) to the PASCAL computer language. The inflexibility of PASCAL eventually led us to convert GELLAB-I to the portable C/UNIX/X-windows environment called GELLAB-II [42, 43, 44, 45] The DECsystem-10 SAIL version, GELLAB-I, was exported to research labs at Univ. of Chicago (Eric Lester) and Univ. of Kiel (Heinz Busse). The Unix version, GELLAB-II, was exported to a number of research labs around the world (CDC with Jim Myrick [44], Univ. Zurich with Peter Sonderegger, Agr. Univ. Norway with Trygve Krekling, and others) and led to a commercial subset version for Windows PCs called GELLAB-II++ by CSPI/Scanalytics.
...
Anchor | ||||
---|---|---|---|---|
|
Anchor | ||||
---|---|---|---|---|
|
...