Investigation of the usefulness of liver-specific deconvolution method by establishing a liver benchmark dataset.

Immune responses in the liver are related to the development and progression of liver failure, and precise prediction of their behavior is important. Deconvolution is a methodology for estimating the immune cell proportions from the transcriptome, and it is mainly applied to blood-derived samples and tumor tissues. However, the influence of tissue-specific modeling on the estimation results has rarely been investigated. Here, we constructed a system to evaluate the performance of the deconvolution method on liver transcriptome data. We prepared seven mouse liver injury models using small-molecule compounds and established a benchmark dataset with corresponding liver bulk RNA-Seq and immune cell proportions. RNA-Seq expression for nine leukocyte subsets and four liver-associated cell types were obtained from the Gene Expression Omnibus to provide a reference. We found that the combination of reference cell sets affects the estimation results of reference-based deconvolution methods and established a liver-specific deconvolution by optimizing the reference cell set for each cell to be estimated. We applied this model to independent datasets and showed that liver-specific modeling is highly extrapolatable. We expect that this approach will enable sophisticated estimation from rich tissue data accumulated in public databases and to obtain information on aggregated immune cell trafficking.

Rat Deconvolution as Knowledge Miner for Immune Cell Trafficking from Toxicogenomics Databases.

Toxicogenomics databases are useful for understanding biological responses in individuals because they include a diverse spectrum of biological responses. Although these databases contain no information regarding immune cells in the liver, which are important in the progression of liver injury, deconvolution that estimates cell-type proportions from bulk transcriptome could extend immune information. However, deconvolution has been mainly applied to humans and mice and less often to rats, which are the main target of toxicogenomics databases. Here, we developed a deconvolution method for rats to retrieve information regarding immune cells from toxicogenomics databases. The rat-specific deconvolution showed high correlations for several types of immune cells between spleen and blood, and between liver treated with toxicants compared with those based on human and mouse data. Additionally, we found 4 clusters of compounds in Open TG-GATEs database based on estimated immune cell trafficking, which are different from those based on transcriptome data itself. The contributions of this work are three-fold. First, we obtained the gene expression profiles of 6 rat immune cells necessary for deconvolution. Second, we clarified the importance of species differences on deconvolution. Third, we retrieved immune cell trafficking from toxicogenomics databases. Accumulated and comparable immune cell profiles of massive data of immune cell trafficking in rats could deepen our understanding of enable us to clarify the relationship between the order and the contribution rate of immune cells, chemokines and cytokines, and pathologies. Ultimately, these findings will lead to the evaluation of organ responses in Adverse Outcome Pathway.

[Understanding of Multiple Effects of Low Molecular Weight Compounds with Factor Analysis].

The effects of drugs and other low-molecular-weight compounds are complex and may be unintended by the developer. These compounds and drugs should be avoided if these unintended effects are harmful; however, unintended effects are not always as harmful as suggested by drug repositioning. Therefore, a comprehensive understanding of complex drug actions is essential. Omics data can be regarded as the nonarbitrary transformation of biological information about a sample into comprehensive numerical information comprising multivariate data with a large number of variables. However, the changes are often based on a small number of elements in different dimensions (i.e., latent variables). The omics data of compound-treated samples comprehensively capture the complex effects of compounds, including their unrecognized aspects. Therefore, finding latent variables in these data is expected to contribute to the understanding of multiple effects. In particular, it can be interpreted as decomposing multiple effects into a smaller number of easily understandable effects. Although latent variable models of omics data have been used to understand the mechanisms of diseases, no approach has considered the multiple effects of compounds and their decomposition. Therefore, we propose to decompose and understand the multiple effects of low-molecular-weight compounds without arbitrariness and have been developing analytical methods and verifying their usefulness. In particular, we focused on classical factor analysis among latent variable models and have been examining the biological validity of the estimates obtained under linear assumptions.

Investigation of a Data Split Strategy Involving the Time Axis in Adverse Event Prediction Using Machine Learning.

Adverse events are a serious issue in drug development, and many prediction methods using machine learning have been developed. The random split cross-validation is the de facto standard for model building and evaluation in machine learning, but care should be taken in adverse event prediction because this approach does not strictly match the real-world situation. The time split, which uses the time axis, is considered suitable for real-world prediction. However, the differences in model performance obtained using the time and random splits are not clear due to the lack of comparable studies. To understand the differences, we compared the model performance between the time and random splits using nine types of compound information as input, eight adverse events as targets, and six machine learning algorithms. The random split showed higher area under the curve values than did the time split for six of eight targets. The chemical spaces of the training and test datasets of the time split were similar, suggesting that the concept of applicability domain is insufficient to explain the differences derived from the splitting. The area under the curve differences were smaller for the protein interaction than for the other datasets. Subsequent detailed analyses suggested the danger of confounding in the use of knowledge-based information in the time split. These findings indicate the importance of understanding the differences between the time and random splits in adverse event prediction and suggest that appropriate use of the splitting strategies and interpretation of results are necessary for the real-world prediction of adverse events. We provide the analysis code and datasets used in the present study at https://github.com/mizuno-group/AE_prediction.

Decomposition Profile Data Analysis for Deep Understanding of Multiple Effects of Natural Products.

It is difficult to understand the entire effect of a natural product because such products generally have multiple effects. We propose a strategy to understand these effects effectively by decomposing them with a profile data analysis method we developed. A transcriptome profile data set was obtained from a public database and analyzed. Considering their high similarity in structure and transcriptome profile, we focused on rescinnamine and syrosingopine. Decomposed effects predicted clear differences between the compounds. Two of the decomposed effects, SREBF1 activation and HDAC inhibition, were investigated experimentally because the relationship between these effects and the compounds had not yet been reported. Analyses in vitro validated these effects, and their strength was consistent with predicted scores. Moreover, the number of outliers in decomposed effects per compound was higher in natural products than in drugs in the data set, which is consistent with the nature of the effects of natural products.

Interesting Properties of Profile Data Analysis in the Understanding and Utilization of the Effects of Drugs.

Profile data is defined as data which describes the properties of an object. Omics data of a specimen is profile data because its comprehensiveness supports the idea that omics data is numeric information which reflects biological information of the specimen. In general, omics data analysis utilizes an existing body of biological knowledge, while some profile data analysis methods are independent of existing knowledge, which is suitable for uncovering unidentified aspects of a specimen of interest. The effects of a small compound, such as drugs, are multiple, and include unrecognized effects, even by the developers. To uncover such unrecognized effects, it is useful to employ profile data analysis independent of existing knowledge. In this review, we summarize what profile data is, properties of profile data analysis, and current applications of profile data in order to understand and utilize the effects of small compounds, in particular, in a recently developed method to decompose multiple effects of a drug.

Decomposition profile data analysis of multiple drug effects identifies endoplasmic reticulum stress-inducing ability as an unrecognized factor.

Chemicals have multiple effects in biological systems. Because their on-target effects dominate the output, their off-target effects are often overlooked and can sometimes cause dangerous adverse events. Recently, we developed a novel decomposition profile data analysis method, orthogonal linear separation analysis (OLSA), to analyse multiple effects. In this study, we tested whether OLSA identified the ability of drugs to induce endoplasmic reticulum (ER) stress as a previously unrecognized factor. After analysing the transcriptome profiles of MCF7 cells treated with different chemicals, we focused on a vector characterized by well-known ER stress inducers, such as ciclosporin A. We selected five drugs predicted to be unrecognized ER stress inducers, based on their inducing ability scores derived from OLSA. These drugs actually induced X-box binding protein 1 splicing, an indicator of ER stress, in MCF7 cells in a concentration-dependent manner. Two structurally different representatives of the five test compounds exhibited similar results in HepG2 and HuH7 cells, but not in PXB primary hepatocytes derived from human-liver chimeric mice. These results indicate that our decomposition strategy using OLSA uncovered the ER stress-inducing ability of drugs as an unrecognized effect, the manifestation of which depended on the background of the cells.