Liang Wei's task force reported a large-scale tumor drug protein perturbation effect screening data set

analysis of the association between genotype and phenotype at the genotype and cellular levels specified by evolution is a core topic in molecular biology research. In terms of methodology, the genetic perturbation experiment and the genotype-expression characteristic statistical analysis are two dominant pathways that can reveal either causal or correlative regulator-target logic, respectively. Considering that the search space for both genotype and manifestation is very large, gene disturbance experiments are often confined to a single regulator or a specific target.
In contrast, the excellent extension of statistical association analysis has become a common means of large-scale interpretation of genotype-expressional connections in the context of rapid accumulation of multi-histological data, and classic cases include genome-wide association analysis (GWAS), expression quantitative genotype base analysis (eQTL), allegen-specific expression analysis (ASE), and gene co-expression network analysis (co-expression network). However, the statistical association analysis paradigm is susceptible to factors such as the upper role (epistatic effect), thus omitting genotypes with direct logical associations, d?e.- dmissive pairings.
recent years, with the rise of CRISPR gene editing technology, single-cell sequencing technology and high-volume image processing technology, it has become possible to disturb genotypes on a large scale and quickly capture
-degree expressive information. Among them, the field of cancer research is one of the biggest beneficiaries of the perturbation biology research paradigm. Specifically, by perturbing the sequence, expression, positioning, or modification of genes, the experimenters then measured the characteristics of RNA, protein expression, or cell state as a reaction, and finally discovered the relationship between disease, genes, and treatment.
The Cancer Dependency Map Program, led by the Broad Institute in the United States, screened the effects of genome-wide single gene knockout on cell proliferation effects in thousands of cancer cell linees, while the L1000 program (formerly known as Connectivity Map) database has more than a million small molecular compounds, gene over-expressions, or gene knocks and other perturbation processes.
However, existing large-scale gene or compound screening data sets lack response indicators centered on protein expression levels, mainly because large-scale quantification of proteins is still difficult, so applying applications such as quantitative proteomics to large-scale perturbation biology research is not a viable path. Considering that protein is the main carrier of cellular function and the direct target of most anti-tumor drugs, the response of expression level and modification state to external stimulation is a key index to reflect the change of cell state and an important clue to anti-tumor mechanism. Therefore, how to integrate protein level monitoring into large-scale cancer disturbance biology research is an urgent problem to be solved.
November 6, 2020, the MD Anderson Cancer Center Liang Wei Group and the Oregon Health and Science University Gordon Mills Task Force published a paper in the journal Lancer Cell entitled Large-scale characterization of drug treatments of clinically relevant proteins in cancer cell lines.
the study first reported a large-scale anti-tumor drug effect screening data set using protein expression levels as a perturbation indicator. This work was carried out by the team members Dr. Zhao Wei, Dr. Li Jun and Dr. Chen Meiru.
Compared to quantitative proteomics, reverse protein microarray (RPPA) offers the advantages of high sample pass, high sensitivity, low sample demand, and low cost, making it suitable for application to large-scale expression level detection of specific functional protein groups. Based on this technique, researchers at MD Anderson Cancer Center have previously accurately quantified more than 8,000 patient samples from 32 cancer types and hundreds of cancer-related proteins in more than 600 cell line systems from 32 cancer types in the Cancer Genome Map (TCGA).
In the study, the authors systematically described the cell state disturbance effects of about 170 preclinical or clinically applied drugs on nearly 320 cancer cell line, using the expression levels of 210 cancer-related proteins measured by RPPA. After rigorous quality verification based on internal repeat group comparisons, the authors were given more than 15,000 perturbation experimental maps. This dataset also exhibits good cross-platform repeatability in subsequent comparisons with externally disclosed data such as CCLE quantitative proteomics and L1000. It is worth mentioning that the data set also includes the determination of effects under different gradients and time processes of pairing the same drug with cancer cell line, resulting in
-degree data types that go beyond single dose, single point in time, and rich biological significance derived from it.
this data set, the authors focused on the association between the resistance strength of cancer cells provided by cancer cell line sensitivity database GDSC and its own protein expression spectrum characteristics. A key finding is that, although, as has been revealed in previous studies, static protein expression data from cancer cells at untreated time can be better used to predict their response to drugs that target specific signaling path paths, accuracy can be significantly improved when drug-treated dynamic protein expression data are added to the predictive model. The results showed the advantages of perturbation biology techniques over statistical association analysis in connecting genotypes (cancer cell protein expression spectrum) and spectral (resistance) from the perspective of cancer cell-drug interoperability.
high dimensional and uniform characteristics of the data set make it possible to establish a large-scale interoperability network of proteins formed according to the classification of drug effects and protein function. Analyzing this network, the authors found that drugs known to have similar biological targets tend to be grouped together, while protein groups with significantly disturbed expression levels from the same drug have significantly more known interconnects. In addition, when the drug-protein network and cancer cell resistance data were analyzed in combination, the authors were able to predict a combination of anticancer drugs that might have potential binding effects by the reverse perturbation of the corresponding signaling path. Interestingly, several of these combinations have been present in specific studies or clinical trials in the past, thus demonstrating the great potential of the determination of large-scale drug protein disturbance effects in guiding clinical drug use.
In summary, for the first time, the data set achieved protein expression as a benchmark in large-scale cancer drug effect screening, making up for significant deficiencies in the interpretation of drug anti-tumor mechanisms and effect prediction due to the lack of protein expression information. In addition, considering that the vast majority of commonly used cancer cell systems already have uniform, high-quality multi-histological data, including RNA, protein, DNA methylation, miRNA, gene mutations, drug sensitivity, etc., the protein perturbation data provided by this study will be able to perform value beyond the exponential level of their own data content through multimodal integration analysis, providing an excellent opportunity to clarify the cross-effects of the drug response of cancer cells at various molecular levels. (Source: Science Network Ikai)
related paper information: