Evaluating the molecule-based prediction of clinical drug responses in cancer

Ding, Zijian; Zu, Songpeng; Gu, Jin

doi:10.1093/bioinformatics/btw344

Cited by 133 publications

(118 citation statements)

References 34 publications

Supporting

Mentioning

117

Contrasting

Unclassified

Order By: Relevance

“…Molecular changes induced by treatment precede changes in morphology or these biomarkers, and may provide proximal endpoints of drug response40. The optical OSI from dual-mode images using metal complex-based probes captures these drug-induced changes in GSH and ROS, which have been hypothesized to correlate closely with treatment outcomes641.…”

Section: Discussionmentioning

confidence: 99%

Two-photon dual imaging platform for in vivo monitoring cellular oxidative stress in liver injury

Wang

Zhang

Bridle

et al. 2017

Sci Rep

View full text Add to dashboard Cite

Oxidative stress reflects an imbalance between reactive oxygen species (ROS) and antioxidants, which has been reported as an early unifying event in the development and progression of various diseases and as a direct and mechanistic indicator of treatment response. However, highly reactive and short-lived nature of ROS and antioxidant limited conventional detection agents, which are influenced by many interfering factors. Here, we present a two-photon sensing platform for in vivo dual imaging of oxidative stress at the single cell-level resolution. This sensing platform consists of three probes, which combine the turn-on fluorescent transition-metal complex with different specific responsive groups for glutathione (GSH), hydrogen peroxide (H2O2) and hypochlorous acid (HOCl). By combining fluorescence intensity imaging and fluorescence lifetime imaging, these probes totally remove any possibility of crosstalk from in vivo environmental or instrumental factors, and enable accurate localization and measurement of the changes in ROS and GSH within the liver. This precedes changes in conventional biochemical and histological assessments in two distinct experimental murine models of liver injury. The ability to monitor real-time cellular oxidative stress with dual-modality imaging has significant implications for high-accurate, spatially configured and quantitative assessment of metabolic status and drug response.

show abstract

Section: Discussionmentioning

confidence: 99%

Two-photon dual imaging platform for in vivo monitoring cellular oxidative stress in liver injury

Wang

Zhang

Bridle

et al. 2017

Sci Rep

View full text Add to dashboard Cite

show abstract

“…For the GDSC dataset, raw gene expression data were downloaded from ArrayExpress (E-MTAB-3610) and response outcomes from https:/www.cancerrxgene.org release 7.0. Gene expression data of TCGA patients were downloaded from the Firehose Broad GDAC (version published on 28.01.2016) and the response outcome was obtained from (Ding et al, 2016). Patient datasets from clinical trials were obtained from the Gene Expression Omnibus (GEO), and the PDX dataset was obtained from the supplementary material of (Gao et al, 2015).…”

Section: Datasetsmentioning

confidence: 99%

“…Finally, each network was re-trained on the selected settings using the train and validation data together for each drug. We used Adagrad for optimizing the parameters of AITL and the baselines (Iorio et al, 2016) cell line Bortezomib targeted source 391 11609 GSE18864 (Silver et al, 2010) clinical trial Cisplatin Chemotherapy target 24 11768 GSE23554 (Marchion et al, 2011) clinical trial Cisplatin Chemotherapy target 28 11768 TCGA (Ding et al, 2016) patient Cisplatin Chemotherapy target 66 11768 GDSC (Iorio et al, 2016) cell line Cisplatin Chemotherapy source 829 11768 GSE25065 (Hatzis et al, 2011) clinical trial Docetaxel Chemotherapy target 49 8119 GSE28796 (Lehmann et al, 2011) clinical trial Docetaxel Chemotherapy target 12 8119 GSE6434 (Chang et al, 2005) clinical trial Docetaxel Chemotherapy target 24 8119 TCGA (Ding et al, 2016) patient Docetaxel Chemotherapy target 16 8119 GDSC (Iorio et al, 2016) cell line Docetaxel Chemotherapy source 829 8119 GSE15622 (Ahmed et al, 2007) clinical trial Paclitaxel Chemotherapy target 20 11731 GSE22513 (Bauer et al, 2010) clinical trial Paclitaxel Chemotherapy target 14 11731 GSE25065 (Hatzis et al, 2011) clinical trial Paclitaxel Chemotherapy target 84 11731 PDX (Gao et al, 2015) animal (mouse) Paclitaxel Chemotherapy target 43 11731 TCGA (Ding et al, 2016) patient Paclitaxel Chemotherapy target 35 11731 GDSC (Iorio et al, 2016) cell line Paclitaxel Chemotherapy source 389 11731 * Number of genes in common between the source and all of the target data for each drug (Duchi et al, 2011) implemented in the PyTorch framework, except for the method of (Geeleher et al, 2014) which was implemented in R. We used the author's implementations for the method of (Geeleher et al, 2014), MOLI, PRECISE, and ProtoNet. For ADDA, we used an existing implementation from https://github.com/jvanvugt/ pytorch-domain-adaptation, and we implemented the method of (Chen et al, 2017) from scratch.…”

Section: Does Aitl Outperform a Baseline For Inductive Transfer Learnmentioning

confidence: 99%

AITL: Adversarial Inductive Transfer Learning with input and output space adaptation for pharmacogenomics

Noghabi

Peng

Zolotareva

et al. 2020

Preprint

View full text Add to dashboard Cite

Motivation: the goal of pharmacogenomics is to predict drug response in patients using their singleor multi-omics data. A major challenge is that clinical data (i.e. patients) with drug response outcome is very limited, creating a need for transfer learning to bridge the gap between large pre-clinical pharmacogenomics datasets (e.g. cancer cell lines), as a source domain, and clinical datasets as a target domain. Two major discrepancies exist between pre-clinical and clinical datasets: 1) in the input space, the gene expression data due to difference in the basic biology, and 2) in the output space, the different measures of the drug response. Therefore, training a computational model on cell lines and testing it on patients violates the i.i.d assumption that train and test data are from the same distribution. Results: We propose Adversarial Inductive Transfer Learning (AITL), a deep neural network method for addressing discrepancies in input and output space between the pre-clinical and clinical datasets. AITL takes gene expression of patients and cell lines as the input, employs adversarial domain adaptation and multi-task learning to address these discrepancies, and predicts the drug response as the output. To the best of our knowledge, AITL is the first adversarial inductive transfer learning method to address both input and output discrepancies. Experimental results indicate that AITL outperforms state-of-the-art pharmacogenomics and transfer learning baselines and may guide precision oncology more accurately. Availability of codes and supplementary material: https://github.com/hosseinshn/AITL Contact: ccollins@prostatecentre.com and ester@cs.sfu.ca © 1

show abstract

“…• Genomics of Drug Sensitivity in Cancer (GDSC) cell lines dataset (Iorio et al, 2016) • Patient-Derived Xenograft (PDX) Encyclopedia dataset (Gao et al, 2015) • TCGA patients with the drug response available in their records (Ding et al, 2016) • TCGA patients without the drug response (Weinstein et al, 2013) The GDSC dataset (Yang et al, 2012;Iorio et al, 2016) has created a multi-omics dataset of more than a thousand cell lines from different cancer types, screened with 265 targeted and chemotherapy drugs. We use GDSC as the training dataset due to a high number of screened drugs.…”

Section: Datasetsmentioning

confidence: 99%

“…For in silico drug response prediction, translatability in the simplest case means that a model with good performance (e.g., high prediction accuracy) on in vitro datatrained on more samples compared to in vivo data-should also have good performance on in vivo data. The majority of studies suggest that gene expression data are the most effective data type for drug response prediction (Iorio et al, 2016;Geeleher et al, 2014;Ding et al, 2016;Graim et al, 2018). Geeleher et al (Geeleher et al, 2014) showed that a ridge regression model trained on GDCS gene expression data is translatable to Docetaxel, Cisplatin, Erlotinib, and Bortezomib clinical trial data.…”

Section: Introductionmentioning

confidence: 99%

MOLI: Multi-Omics Late Integration with deep neural networks for drug response prediction

Sharifi-Noghabi

Zolotareva

Ester

2019

Preprint

View full text Add to dashboard Cite

Motivation: Historically, gene expression has been shown to be the most informative data for drug response prediction. Recent evidence suggests that integrating additional omics can improve the prediction accuracy which raises the question of how to integrate the additional omics. Regardless of the integration strategy, clinical utility and translatability are crucial. Thus, we reasoned a multi-omics approach combined with clinical datasets would improve drug response prediction and clinical relevance. Results: We propose MOLI, a Multi-Omics Late Integration method based on deep neural networks. MOLI takes somatic mutation, copy number aberration, and gene expression data as input, and integrates them for drug response prediction. MOLI uses type-specific encoding subnetworks to learn features for each omics type, concatenates them into one representation and optimizes this representation via a combined cost function consisting of a triplet loss and a binary cross-entropy loss. The former makes the representations of responder samples more similar to each and different from the non-responders, and the latter makes this representation predictive of the response values. We validate MOLI on in vitro and in vivo datasets for five chemotherapy agents and two targeted therapeutics. Compared to state-of-the-art single-omics and early integration multi-omics methods, MOLI achieves higher prediction accuracy in external validations. Moreover, a significant improvement in MOLI's performance is observed for targeted drugs when training on a pan-drug input, i.e. using all the drugs with the same target compared to training only on drug-specific inputs. MOLI's high predictive power suggests it may have utility in precision oncology. Availability of the implemented codes: https://github.com/hosseinshn/MOLI Contact: ccollins@prostatecentre.com and

show abstract

Evaluating the molecule-based prediction of clinical drug responses in cancer

Abstract: Supplementary data are available at Bioinformatics online.

Cited by 133 publications

References 34 publications

Two-photon dual imaging platform for in vivo monitoring cellular oxidative stress in liver injury

Two-photon dual imaging platform for in vivo monitoring cellular oxidative stress in liver injury

AITL: Adversarial Inductive Transfer Learning with input and output space adaptation for pharmacogenomics

MOLI: Multi-Omics Late Integration with deep neural networks for drug response prediction

Contact Info

Product

Resources

About