Towards a knowledge-based human protein atlas

M Uhlen, P Oksvold, L Fagerberg, E Lundberg… - Nature …, 2010 - nature.com
M Uhlen, P Oksvold, L Fagerberg, E Lundberg, K Jonasson, M Forsberg, M Zwahlen…
Nature biotechnology, 2010nature.com
NATURE BIoTECHNoloGy volume 28 number 12 DeCember 2010 1249 the resulting
annotated protein expression. The annotated protein expression in the various cell types are
obtained by a knowledge-based process taking into account staining from at least two
antibodies, literature and additional experimental data. For the ESR1 (Fig. 2), one of the
antibodies shows weak staining in tissues such as those of the prostate and urinary bladder,
most likely due to nonspecific staining, but the conflation of data from the three antibodies …
NATURE BIoTECHNoloGy volume 28 number 12 DeCember 2010 1249 the resulting annotated protein expression. The annotated protein expression in the various cell types are obtained by a knowledge-based process taking into account staining from at least two antibodies, literature and additional experimental data. For the ESR1 (Fig. 2), one of the antibodies shows weak staining in tissues such as those of the prostate and urinary bladder, most likely due to nonspecific staining, but the conflation of data from the three antibodies suggests exclusive expression in breast and female reproductive tissues. It is important to point out that this annotated protein expression should be considered as the best estimate of the true expression based on the experimental data available, but that additional data and input from the scientific community may lead to subsequent revisions for a small fraction of the human genes, these subcellular localization data are a valuable resource for defining the protein content of the various compartments of the human cell and can thus constitute a starting point for further in-depth functional studies. A new organ view has been designed to allow cell types of similar origin to be easily compared. The tissues and organs have been divided into 12 functional classes, including central nervous system (CNS), hematopoietic, cardiovascular and female and male reproductive tissues. In this release, a total of 66 normal cell types from 46 tissues and organs have been scored. As an example, in Figure 2, we show part of the protein profiles for the human estrogen receptor 1 (ESR1), including the results from antibody staining of three independent antibodies and or transcript level and proteins inferred from homology6 (Fig. 1). The chromosomal coverage of protein-coding genes is shown in Figure 1a and the status for a selection of important protein classes is reported in Figure 1b. Almost 80% of the human kinases and Src-homology 2 domain–containing proteins and> 50% of the transcription factors have protein profiling data in the atlas. We introduce the concept of annotated protein expression for paired antibodies, in which two or more independent antibodies are used to validate the staining pattern of each other. The immunohistochemical staining in each tissue or organ by the independent antibodies is compared and a new annotated protein expression is manually curated for each cell type in each tissue or organ. A reliability score is generated as an estimation of the degree of knowledge-based certainty of the reported expression profile. The reliability score is based on the similarity of the staining for the different antibodies, but also takes into account available information from literature, bioinformatics predictions and additional experimental evidence, such as western blots, transcript profiling and/or small interfering RNA knockdowns
nature.com