MassExplorer: a computational tool for analyzing desorption electrospray ionization mass spectrometry data
Bioinformatics (Oxford, England)
High-throughput gene expression can be used to address a wide range of fundamental biological problems, but datasets of an appropriate size are often unavailable. Moreover, existing transcriptomics simulators have been criticised because they fail to emulate key properties of gene expression data. In this paper, we develop a method based on a conditional generative adversarial network to generate realistic transcriptomics data for E. coli and humans. We assess the performance of our approach across several tissues and cancer types.We show that our model preserves several gene expression properties significantly better than widely used simulators such as SynTReN or GeneNetWeaver. The synthetic data preserves tissue and cancer-specific properties of transcriptomics data. Moreover, it exhibits real gene clusters and ontologies both at local and global scales, suggesting that the model learns to approximate the gene expression manifold in a biologically meaningful way.Code is available at: https://github.com/rvinas/adversarial-gene-expression.Supplementary data are available at Bioinformatics online.
View details for DOI 10.1093/bioinformatics/btab282
View details for PubMedID 34009252
In situ DESI-MSI lipidomic profiles of mucosal margin of oral squamous cell carcinoma.
2021; 70: 103529
Although there is consensus that the optimal safe margin is ≥ 5mm, obtaining clear margins (≥5 mm) intraoperatively seems to be the major challenge. We applied a molecular diagnostic method at the lipidomic level to determine the safe surgical resection margin of OSCC by desorption electrospray ionisation mass spectrometry imaging (DESI-MSI).By overlaying mass spectrometry images with hematoxylin-eosin staining (H&E) from 18 recruited OSCC participants, the mass spectra of all pixels across the diagnosed tumour and continuous mucosal margin regions were extracted to serve as the training and validation datasets. A Lasso regression model was used to evaluate the test performance.By leave-one-out validation, the Lasso model achieved 88.6% accuracy in distinguishing between tumour and normal regions. To determine the safe surgical resection distance and margin status of OSCC, a set of 14 lipid ions that gradually decreased from tumour to normal tissue was assigned higher weight coefficients in the Lasso model. The safe surgical resection distance of OSCC was measured using the developed 14 lipid ion molecular diagnostic model for clinical reference. The overall accuracy of predicting tumours, positive margins, and negative margins was 92.6%.The spatial segmentation results based on our diagnostic model not only clearly delineated the tumour and normal tissue, but also distinguished the different status of surgical margins. Meanwhile, the safe surgical resection margin of OSCC on frozen sections can also be accurately measured using the developed diagnostic model.This study was supported by Nanjing Municipal Key Medical Laboratory Constructional Project Funding (since 2016) and the Centre of Nanjing Clinical Medicine Tumour (since 2014).
View details for DOI 10.1016/j.ebiom.2021.103529
View details for PubMedID 34391097
Oral squamous cell carcinoma diagnosed from saliva metabolic profiling.
Proceedings of the National Academy of Sciences of the United States of America
Saliva is a noninvasive biofluid that can contain metabolite signatures of oral squamous cell carcinoma (OSCC). Conductive polymer spray ionization mass spectrometry (CPSI-MS) is employed to record a wide range of metabolite species within a few seconds, making this technique appealing as a point-of-care method for the early detection of OSCC. Saliva samples from 373 volunteers, 124 who are healthy, 124 who have premalignant lesions, and 125 who are OSCC patients, were collected for discovering and validating dysregulated metabolites and determining altered metabolic pathways. Metabolite markers were reconfirmed at the primary tissue level by desorption electrospray ionization MS imaging (DESI-MSI), demonstrating the reliability of diagnoses based on saliva metabolomics. With the aid of machine learning (ML), OSCC and premalignant lesions can be distinguished from the normal physical condition in real time with an accuracy of 86.7%, on a person by person basis. These results suggest that the combination of CPSI-MS and ML is a feasible tool for accurate, automated diagnosis of OSCC in clinical practice.
View details for DOI 10.1073/pnas.2001395117
View details for PubMedID 32601197
Metabolite therapy guided by liquid biopsy proteomics delays retinal neurodegeneration.
2020; 52: 102636
Neurodegenerative diseases are incurable disorders caused by progressive neuronal cell death. Retinitis pigmentosa (RP) is a blinding neurodegenerative disease that results in photoreceptor death and progresses to the loss of the entire retinal network. We previously found that proteomic analysis of the adjacent vitreous served as way to indirectly biopsy the retina and identify changes in the retinal proteome.We analyzed protein expression in liquid vitreous biopsies from autosomal recessive (ar)RP patients with PDE6A mutations and arRP mice with Pde6ɑ mutations. Proteomic analysis of retina and vitreous samples identified molecular pathways affected at the onset of photoreceptor death. Based on affected molecular pathways, arRP mice were treated with a ketogenic diet or metabolites involved in fatty-acid synthesis, oxidative phosphorylation, and the tricarboxylic acid (TCA) cycle.Dietary supplementation of a single metabolite, ɑ-ketoglutarate, increased docosahexaeonic acid levels, provided neuroprotection, and enhanced visual function in arRP mice. A ketogenic diet delayed photoreceptor cell loss, while vitamin B supplementation had a limited effect. Finally, desorption electrospray ionization mass spectrometry imaging (DESI-MSI) on ɑ-ketoglutarate-treated mice revealed restoration of metabolites that correlated with our proteomic findings: uridine, dihydrouridine, and thymidine (pyrimidine and purine metabolism), glutamine and glutamate (glutamine/glutamate conversion), and succinic and aconitic acid (TCA cycle).This study demonstrates that replenishing TCA cycle metabolites via oral supplementation prolongs retinal function and provides a neuroprotective effect on the photoreceptor cells and inner retinal network.NIH grants [R01EY026682, R01EY024665, R01EY025225, R01EY024698, R21AG050437, P30EY026877, 5P30EY019007, R01EY018213, F30EYE027986, T32GM007337, 5P30CA013696], NSF grant CHE-1734082.
View details for DOI 10.1016/j.ebiom.2020.102636
View details for PubMedID 32028070
- Developing Successful Mental Illness Prevention Efforts on University Campuses: A Local Look Based on the CARES Program at Stanford University Journal of the American Academy of Child and Adolescent Psychiatry (JAACAP) Connect 2020
Identification of Diagnostic Metabolic Signatures in Clear Cell Renal Cell Carcinoma Using Mass Spectrometry Imaging.
International journal of cancer
Clear cell renal cell carcinoma (ccRCC) is the most common and lethal subtype of kidney cancer. Intraoperative frozen section (IFS) analysis is used to confirm the diagnosis during partial nephrectomy (PN). However, surgical margin evaluation using IFS analysis is time consuming and unreliable, leading to relatively low utilization. In this study, we demonstrated the use of desorption electrospray ionization mass spectrometry imaging (DESI-MSI) as a molecular diagnostic and prognostic tool for ccRCC. DESI-MSI was conducted on fresh-frozen 23 normal-tumor paired nephrectomy specimens of ccRCC. An independent validation cohort of 17 normal-tumor pairs were analyzed. DESI-MSI provides two-dimensional molecular images of tissues with mass spectra representing small metabolites, fatty acids, and lipids. These tissues were subjected to histopathologic evaluation. A set of metabolites that distinguish ccRCC from normal kidney were identified by performing least absolute shrinkage and selection operator (Lasso) and log-ratio Lasso analysis. Lasso analysis with leave-one-patient-out cross validation selected 57 peaks from over 27,000 metabolic features across 37,608 pixels obtained using DESI-MSI of ccRCC and normal tissues. Baseline Lasso of metabolites predicted the class of each tissue to be normal or cancerous tissue with an accuracy of 94% and 76%, respectively. Combining the baseline Lasso with the ratio of glucose to arachidonic acid could potentially reduce scan time and improve accuracy to identify normal (82%) and ccRCC (88%) tissue. DESI-MSI allows rapid detection of metabolites associated with normal and ccRCC with high accuracy. As this technology advances, it could be used for rapid intraoperative assessment of surgical margin status. This article is protected by copyright. All rights reserved.
View details for DOI 10.1002/ijc.32843
View details for PubMedID 31863456
Using DESI-MSI to identify the genetic basis and tumorigenic mechanism of pheochromocytomas
AMER CHEMICAL SOC. 2019
View details for Web of Science ID 000525055501165
Can machine learning be used to learn laws of natural science? Illustration for Planck's blackbody radiation
AMER CHEMICAL SOC. 2019
View details for Web of Science ID 000525055501166
The 3D Structure of Human DP Prostaglandin G-Protein-Coupled Receptor Bound to Cyclopentanoindole Antagonist, Predicted Using the DuplexBiHelix Modification of the GEnSeMBLE Method.
Journal of chemical theory and computation
2018; 14 (3): 1624–42
Prostaglandins play a critical physiological role in both cardiovascular and immune systems, acting through their interactions with 9 prostanoid G protein-coupled receptors (GPCRs). These receptors are important therapeutic targets for a variety of diseases including arthritis, allergies, type 2 diabetes, and cancer. The DP prostaglandin receptor is of interest because it has unique structural and physiological properties. Most notably, DP does not have the 3-6 ionic lock common to Class A GPCRs. However, the lack of X-ray structures for any of the 9 prostaglandin GPCRs hampers the application of structure-based drug design methods to develop more selective and active medications to specific receptors. We predict here 3D structures for the DP prostaglandin GPCR, based on the GEnSeMBLE complete sampling with hierarchical scoring (CS-HS) methodology. This involves evaluating the energy of 13 trillion packings to finally select the best 20 that are stable enough to be relevant for binding to antagonists, agonists, and modulators. To validate the predicted structures, we predict the binding site for the Merck cyclopentanoindole (CPI) selective antagonist docked to DP. We find that the CPI binds vertically in the 1-2-7 binding pocket, interacting favorably with residues R3107.40 and K762.54 with additional interactions with S3137.43, S3167.46, S191.35, etc. This binding site differs significantly from that of antagonists to known Class A GPCRs where the ligand binds in the 3-4-5-6 region. We find that the predicted binding site leads to reasonable agreement with experimental Structure-Activity Relationship (SAR). We suggest additional mutation experiments including K762.54, E1293.49, L1233.43, M2706.40, F2746.44 to further validate the structure, function, and activation mechanism of receptors in the prostaglandin family. Our structures and binding sites are largely consistent and improve upon the predictions by Li et al. ( J. Am. Chem. Soc. 2007 , 129 ( 35 ), 10720 ) that used our earlier MembStruk prediction methodology.
View details for DOI 10.1021/acs.jctc.7b00842
View details for PubMedID 29268008
Comparative effectiveness and cost-effectiveness of treat-to-target versus benefit-based tailored treatment of type 2 diabetes in low-income and middle-income countries: a modelling analysis
LANCET DIABETES & ENDOCRINOLOGY
2016; 4 (11): 922-932
Optimal prescription of blood pressure, lipid, and glycaemic control treatments for adults with type 2 diabetes remains unclear. We aimed to compare the effectiveness and cost-effectiveness of two treatment approaches for diabetes management in five low-income and middle-income countries.We developed a microsimulation model to compare a treat-to-target (TTT) strategy, aiming to achieve target levels of biomarkers (blood pressure <130/80 mm Hg, LDL <2·59 mmol/L, and HbA1c <7% [ie, 53·0 mmol/mol]), with a benefit-based tailored treatment (BTT) strategy, aiming to lower estimated risk for complications (to a 10 year cardiovascular risk <10% and lifetime microvascular risk <5%) on the basis of age, sex, and biomarker values. Data were obtained from cohorts in China, Ghana, India, Mexico, and South Africa to span a spectrum of risk profiles.The TTT strategy recommended treatment to a larger number of people-who were generally at lower risk of diabetes complications-than the BTT. The BTT strategy recommended treatment to fewer people at higher risk. Compared with the TTT strategy, the BTT strategy would be expected to avert 24·4-30·5% more complications and be more cost-effective from a societal perspective (saving US$4·0-300·0 per disability-adjusted life-year averted in the countries simulated). Alternative treatment thresholds, matched by total cost or population size treated, did not change the comparative superiority of the BTT strategy, nor did titrating treatment using fasting plasma glucose (for areas without HbA1c testing). However, if insulin were unavailable, the BTT strategy would no longer be superior for preventing microvascular events and was superior only for preventing cardiovascular events.A BTT strategy is more effective and cost-effective than a TTT strategy in low-income and middle-income countries for prevention of both cardiovascular and microvascular complications of type 2 diabetes. However, the superiority of the BTT strategy for averting microvascular complications is contingent on insulin availability.Rosenkranz Prize for Healthcare Research in Developing Countries and US National Institutes of Health (U54 MD010724, DP2 MD010478).
View details for DOI 10.1016/S2213-8587(16)30270-4
View details for Web of Science ID 000393025500028
View details for PubMedID 27717768