DRLM: A Robust Drug Representation Learning Method and Its Applications.

Learning representations from data is a fundamental step for machine learning. High-quality and robust drug representations can broaden the understanding of pharmacology, and improve the modeling of multiple drug-related prediction tasks, which further facilitates drug development. Although there are a number of models developed for drug representation learning from various data sources, few researches extract drug representations from gene expression profiles. Since gene expression profiles of drug-treated cells are widely used in clinical diagnosis and therapy, it is believed that leveraging them to eliminate cell specificity can promote drug representation learning. In this paper, we propose a three-stage deep learning method for drug representation learning, named DRLM, which integrates gene expression profiles of drug-related cells and the therapeutic use information of drugs. Firstly, we construct a stacked autoencoder to learn low-dimensional compact drug representations. Secondly, we utilize an iterative clustering module to reduce the negative effects of cell specificity and noise in gene expression profiles on the low-dimensional drug representations. Thirdly, a therapeutic use discriminator is designed to incorporate therapeutic use information into the drug representations. The visualization analysis of drug representations demonstrates DRLM can reduce cell specificity and integrate therapeutic use information effectively. Extensive experiments on three types of prediction tasks are conducted based on different drug representations, and they show that the drug representations learned by DRLM outperform other representations in terms of most metrics. The ablation analysis also demonstrates DRLM's effectiveness of merging the gene expression profiles with the therapeutic use information. Furthermore, we input the learned representations into the machine learning models for case studies, which indicates its potential to discover new drug-related relationships in various tasks.

Optimization of scan parameters to reduce acquisition time for RESOLVE-based diffusion kurtosis imaging (DKI) in nasopharyngeal carcinoma (NPC).

OBJECTIVE: To shorten acquisition time of readout segmentation of long variable echo trains (RESOLVE)-based diffusion kurtosis imaging (DKI) via Readout Partial Fourier (RPF) and b-value combinations. METHODS: The RESOLVE-based DKI images of 38 patients with nasopharyngeal carcinoma (NPC) were prospectively enrolled. For RESOLVE-based DKI images with 5/8 RPF and without RPF, objective and subjective evaluations of image quality were performed. A total of nine groups with different b-value combinations were simulated, and the influence of different b-value combinations for RESOLVE-RPF-based DKI sequences was assessed using the intraclass correlation coefficient (ICC). RESULTS: The mean values of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) in DKI images without RPF were higher than those with 5/8 RPF (252.9 ± 77.7 vs 247.3 ± 85.5 and 5.8 ± 2.8 vs 5.4 ± 2.3, respectively), but not significantly (p = 0.460 and p = 0.180, respectively). In comparing the ICCs between nine groups of different b-value combinations in RESOLVE-RPF-based DKI, group (200, 800, 2000 s/mm2), group (200, 400, 800, 2000 s/mm2) and group (200, 800, 1500, 2000 s/mm2) were not significantly different (p > 0.001) and showed excellent agreement (0.81-1.00) with that of group (200, 400, 800, 1500, 2000 s/mm2). Using b-value optimization and RPF technology, the group with RPF (200, 400, 800, 2000 s/mm2) showed a 56% reduced scanning compared with the group without RPF (200, 400, 800, 1500, 2000 s/mm2; 3 min 46 s vs 8 min 31 s, respectively). CONCLUSION: DKI with RPF did not significantly affect image quality, but both RPF and different b-value combinations can affect the scanning time. The combination of RPF and b-value optimization can ensure the stability of DKI parameters and reduce the scanning time by 56%. ADVANCES IN KNOWLEDGE: This work is to optimize scan parameters, e.g. RPF and b-value combinations, to reduce acquisition time for RESOLVE-based DKI in NPC. To our knowledge, the effect of RESOLVE-RPF and b-value combinations on DKI has not been reported.

Reproducibility of radiomics features derived from intravoxel incoherent motion diffusion-weighted MRI of cervical cancer.

BACKGROUND: The reproducibility of intravoxel incoherent motion (IVIM)-based radiomics studies in humans has not been reported. PURPOSE: To determine the inter- and intra-observer variability on the reproducibility of IVIM-based radiomics features in cervical cancer (CC). MATERIAL AND METHODS: The IVIM images of 25 patients with CC were retrospectively collected. Based on the high-resolution T2-weighted images, the regions of interest (ROIs) were independently delineated twice in diffusion-weighted images at a b value of 1000 s/mm2 (interval time was one month) by two radiologists. This was done at the largest transversal cross-sections of the tumors. The ROI was subsequently used in apparent diffusion coefficient (ADC), true diffusion coefficient (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f) maps derived from IVIM images. In total, 105 radiomics features were then finally extracted from the IVIM-derived maps. The inter- and intra-observer reproducibility of IVIM-derived features was then evaluated using the intraclass correlation coefficient. RESULTS: Inter- and intra-observer variability affected the reproducibility of radiomics features. D* map had 100% and 95% reproducible features, ADC map had 89% and 93%, D map had 97% and 86%, while f map had 54% and 62% reproducible features with good to excellent reliability in the intra-observer analysis. Similarly, D* map had 90% and 94%, ADC map had 85% and 70%, D map had 81% and 78%, while f map had 41% and 93% reproducible features with good to excellent reliability in the inter-observer analysis. CONCLUSION: Inter- and intra-observer variability can affect radiomics analysis. Cognizant to this, multicenter studies should pay more attention to intra- and inter-observer variability.

Animal-ImputeDB: a comprehensive database with multiple animal reference panels for genotype imputation.

Animal-ImputeDB (http://gong_lab.hzau.edu.cn/Animal_ImputeDB/) is a public database with genomic reference panels of 13 animal species for online genotype imputation, genetic variant search, and free download. Genotype imputation is a process of estimating missing genotypes in terms of the haplotypes and genotypes in a reference panel. It can effectively increase the density of single nucleotide polymorphisms (SNPs) and thus can be widely used in large-scale genome-wide association studies (GWASs) using relatively inexpensive and low-density SNP arrays. However, most animals except humans lack high-quality reference panels, which greatly limits the application of genotype imputation in animals. To overcome this limitation, we developed Animal-ImputeDB, which is dedicated to collecting genotype data and whole-genome resequencing data of nonhuman animals from various studies and databases. A computational pipeline was developed to process different types of raw data to construct reference panels. Finally, 13 high-quality reference panels including ∼400 million SNPs from 2265 samples were constructed. In Animal-ImputeDB, an easy-to-use online tool consisting of two popular imputation tools was designed for the purpose of genotype imputation. Collectively, Animal-ImputeDB serves as an important resource for animal genotype imputation and will greatly facilitate research on animal genomic selection and genetic improvement.