CRISP: a deep learning architecture for GC × GC-TOFMS contour ROI identification, simulation and analysis in imaging metabolomics.

Two-dimensional gas chromatography-time-of-flight mass spectrometry (GC × GC-TOFMS) provides a large amount of molecular information from biological samples. However, the lack of a comprehensive compound library or customizable bioinformatics tool is currently a challenge in GC × GC-TOFMS data analysis. We present an open-source deep learning (DL) software called contour regions of interest (ROI) identification, simulation and untargeted metabolomics profiler (CRISP). CRISP integrates multiple customizable deep neural network architectures for assisting the semi-automated identification of ROIs, contour synthesis, resolution enhancement and classification of GC × GC-TOFMS-based contour images. The approach includes the novel aggregate feature representative contour (AFRC) construction and stacked ROIs. This generates an unbiased contour image dataset that enhances the contrasting characteristics between different test groups and can be suitable for small sample sizes. The utility of the generative models and the accuracy and efficacy of the platform were demonstrated using a dataset of GC × GC-TOFMS contour images from patients with late-stage diabetic nephropathy and healthy control groups. CRISP successfully constructed AFRC images and identified over five ROIs to create a deepstacked dataset. The high fidelity, 512 × 512-pixels generative model was trained as a generator with a Fréchet inception distance of <47.00. The trained classifier achieved an AUROC of >0.96 and a classification accuracy of >95.00% for datasets with and without column bleed. Overall, CRISP demonstrates good potential as a DL-based approach for the rapid analysis of 4-D GC × GC-TOFMS untargeted metabolite profiles by directly implementing contour images. CRISP is available at https://github.com/vivekmathema/GCxGC-CRISP.

Accurate Prediction of Ion Mobility Collision Cross-Section Using Ion's Polarizability and Molecular Mass with Limited Data.

The rotationally averaged collision cross-section (CCS) determined by ion mobility-mass spectrometry (IM-MS) facilitates the identification of various biomolecules. Although machine learning (ML) models have recently emerged as a highly accurate approach for predicting CCS values, they rely on large data sets from various instruments, calibrants, and setups, which can introduce additional errors. In this study, we identified and validated that ion's polarizability and mass-to-charge ratio (m/z) have the most significant predictive power for traveling-wave IM CCS values in relation to other physicochemical properties of ions. Constructed solely based on these two physicochemical properties, our CCS prediction approach demonstrated high accuracy (mean relative error of <3.0%) even when trained with limited data (15 CCS values). Given its ability to excel with limited data, our approach harbors immense potential for constructing a precisely predicted CCS database tailored to each distinct experimental setup. A Python script for CCS prediction using our approach is freely available at https://github.com/MSBSiriraj/SVR_CCSPrediction under the GNU General Public License (GPL) version 3.

GC × GC-TOFMS metabolomics analysis identifies elevated levels of plasma sugars and sugar alcohols in diabetic mellitus patients with kidney failure.

Two dimensional GC (GC × GC)-time-of-flight mass spectrometry (TOFMS) has been used to improve accurate metabolite identification in the chemical industry, but this method has not been applied as readily in biomedical research. Here, we evaluated and validated the performance of high resolution GC × GC-TOFMS against that of GC-TOFMS for metabolomics analysis of two different plasma matrices, from healthy controls (CON) and diabetes mellitus (DM) patients with kidney failure (DM with KF). We found GC × GC-TOFMS outperformed traditional GC-TOFMS in terms of separation performance and metabolite coverage. Several metabolites from both the CON and DM with KF matrices, such as carbohydrates and carbohydrate-conjugate metabolites, were exclusively detected using GC × GC-TOFMS. Additionally, we applied this method to characterize significant metabolites in the DM with KF group, with focused analysis of four metabolite groups: sugars, sugar alcohols, amino acids, and free fatty acids. Our plasma metabolomics results revealed 35 significant metabolites (12 unique and 23 concentration-dependent metabolites) in the DM with KF group, as compared with those in the CON and DM groups (N = 20 for each group). Interestingly, we determined 17 of the 35 (14/17 verified with reference standards) significant metabolites identified from both the analyses were metabolites from the sugar and sugar alcohol groups, with significantly higher concentrations in the DM with KF group than in the CON and DM groups. Enrichment analysis of these 14 metabolites also revealed that alterations in galactose metabolism and the polyol pathway are related to DM with KF. Overall, our application of GC × GC-TOFMS identified key metabolites in complex plasma matrices.

Revisiting chloroplast genomic landscape and annotation towards comparative chloroplast genomes of Rhamnaceae.

BACKGROUND: Massive parallel sequencing technologies have enabled the elucidation of plant phylogenetic relationships from chloroplast genomes at a high pace. These include members of the family Rhamnaceae. The current Rhamnaceae phylogenetic tree is from 13 out of 24 Rhamnaceae chloroplast genomes, and only one chloroplast genome of the genus Ventilago is available. Hence, the phylogenetic relationships in Rhamnaceae remain incomplete, and more representative species are needed. RESULTS: The complete chloroplast genome of Ventilago harmandiana Pierre was outlined using a hybrid assembly of long- and short-read technologies. The accuracy and validity of the final genome were confirmed with PCR amplifications and investigation of coverage depth. Sanger sequencing was used to correct for differences in lengths and nucleotide bases between inverted repeats because of the homopolymers. The phylogenetic trees reconstructed using prevalent methods for phylogenetic inference were topologically similar. The clustering based on codon usage was congruent with the molecular phylogenetic tree. The groups of genera in each tribe were in accordance with tribal classification based on molecular markers. We resolved the phylogenetic relationships among six Hovenia species, three Rhamnus species, and two Ventilago species. Our reconstructed tree provides the most complete and reliable low-level taxonomy to date for the family Rhamnaceae. Similar to other higher plants, the RNA editing mostly resulted in converting serine to leucine. Besides, most genes were subjected to purifying selection. Annotation anomalies, including indel calling errors, unaligned open reading frames of the same gene, inconsistent prediction of intergenic regions, and misannotated genes, were identified in the published chloroplast genomes used in this study. These could be a result of the usual imperfections in computational tools, and/or existing errors in reference genomes. Importantly, these are points of concern with regards to utilizing published chloroplast genomes for comparative genomic analysis. CONCLUSIONS: In summary, we successfully demonstrated the use of comprehensive genomic data, including DNA and amino acid sequences, to build a reliable and high-resolution phylogenetic tree for the family Rhamnaceae. Additionally, our study indicates that the revision of genome annotation before comparative genomic analyses is necessary to prevent the propagation of errors and complications in downstream analysis and interpretation.

Deep learning meets metabolomics: a methodological perspective.

Deep learning (DL), an emerging area of investigation in the fields of machine learning and artificial intelligence, has markedly advanced over the past years. DL techniques are being applied to assist medical professionals and researchers in improving clinical diagnosis, disease prediction and drug discovery. It is expected that DL will help to provide actionable knowledge from a variety of 'big data', including metabolomics data. In this review, we discuss the applicability of DL to metabolomics, while presenting and discussing several examples from recent research. We emphasize the use of DL in tackling bottlenecks in metabolomics data acquisition, processing, metabolite identification, as well as in metabolic phenotyping and biomarker discovery. Finally, we discuss how DL is used in genome-scale metabolic modelling and in interpretation of metabolomics data. The DL-based approaches discussed here may assist computational biologists with the integration, prediction and drawing of statistical inference about biological outcomes, based on metabolomics data.

Traveling Wave Ion Mobility-Derived Collision Cross Section Database for Plant Specialized Metabolites: An Application to Ventilago harmandiana Pierre.

The combination of ion mobility mass spectrometry (IM-MS) and chromatography is a valuable tool for identifying compounds in natural products. In this study, using an ultra-performance liquid chromatography system coupled to a high-resolution quadrupole/traveling wave ion mobility spectrometry/time-of-flight MS (UPLC-TWIMS-QTOF), we have established and validated a comprehensive TWCCSN2 and MS database for 112 plant specialized metabolites. The database included 15 compounds that were isolated and purified in-house and are not commercially available. We obtained accurate m/z, retention times, fragment ions, and TWIMS-derived CCS (TWCCSN2) values for 207 adducts (ESI+ and ESI-). The database included novel 158 TWCCSN2 values from 79 specialized metabolites. In the presence of plant matrix, the CCS measurement was reproducible and robust. Finally, we demonstrated the application of the database to extend the metabolite coverage of Ventilago harmandiana Pierre. In addition to pyranonaphthoquinones, a group of known specialized metabolites in V. harmandiana, we identified flavonoids, xanthone, naphthofuran, and protocatechuic acid for the first time through targeted analysis. Interestingly, further investigation using IM-MS of unknown features suggested the presence of organonitrogen compounds and lipid and lipid-like molecules, which is also reported for the first time. Data are available on the MassIVE (https://massive.ucsd.edu, data set identifier MSV000090213).

Engineering yeast phospholipid metabolism for de novo oleoylethanolamide production.

Phospholipids, the most abundant membrane lipid components, are crucial in maintaining membrane structures and homeostasis for biofunctions. As a structurally diverse and tightly regulated system involved in multiple organelles, phospholipid metabolism is complicated to manipulate. Thus, repurposing phospholipids for lipid-derived chemical production remains unexplored. Herein, we develop a Saccharomyces cerevisiae platform for de novo production of oleoylethanolamide, a phospholipid derivative with promising pharmacological applications in ameliorating lipid dysfunction and neurobehavioral symptoms. Through deregulation of phospholipid metabolism, screening of biosynthetic enzymes, engineering of subcellular trafficking and process optimization, we could produce oleoylethanolamide at a titer of 8,115.7 µg l-1 and a yield on glucose of 405.8 µg g-1. Our work provides a proof-of-concept study for systemically repurposing phospholipid metabolism for conversion towards value-added biological chemicals, and this multi-faceted framework may shed light on tailoring phospholipid metabolism in other microbial hosts.

Metabolic Profiling and Compound-Class Identification Reveal Alterations in Serum Triglyceride Levels in Mice Immunized with Human Vaccine Adjuvant Alum.

Alum has been widely used as an adjuvant for human vaccines; however, the impact of Alum on host metabolism remains largely unknown. Herein, we applied mass spectrometry (MS) (liquid chromatography-MS)-based metabolic and lipid profiling to monitor the effects of the Alum adjuvant on mouse serum at 6, 24, 72, and 168 h post-vaccination. We propose a new strategy termed subclass identification and annotation for metabolomics for class-wise identification of untargeted metabolomics data generated from high-resolution MS. Using this approach, we identified and validated the levels of several lipids in mouse serum that were significantly altered following Alum administration. These lipids showed a biphasic response even 168 h after vaccination. The majority of the lipids were triglycerides (TAGs), where TAGs with long-chain unsaturated fatty acids (FAs) decreased at 24 h and TAGs with short-chain FAs decreased at 168 h. To our knowledge, this is the first report on the impact of human vaccine adjuvant Alum on the host metabolome, which may provide new insights into the mechanism of action of Alum.

Comparison of the metabolic response to over-production of p-coumaric acid in two yeast strains.

The development of robust and efficient cell factories requires understanding of the metabolic changes triggered by the production of the targeted compound. Here we aimed to study how production of p-coumaric acid, a precursor of multiple secondary aromatic metabolites, influences the cellular metabolism of Saccharomyces cerevisiae. We evaluated the growth and p-coumaric acid production in batch and chemostat cultivations and analyzed the transcriptome and intracellular metabolome during steady state in low- and high-producers of p-coumaric acid in two strain backgrounds, S288c or CEN.PK. We found that the same genetic modifications resulted in higher production of p-coumaric acid in the CEN.PK background than in the S288c background. Moreover, the CEN.PK strain was less affected by the genetic engineering as was evident from fewer changes in the transcription profile and intracellular metabolites concentrations. Surprisingly, for both strains we found the largest transcriptional changes in genes involved in transport of amino acids and sugars, which were downregulated. Additionally, in S288c amino acid and protein biosynthesis processes were also affected. We systematically overexpressed or deleted genes with significant transcriptional changes in CEN.PK low and high-producing strains. The knockout of some of the downregulated transporters triggered a 20-50% improvement in the synthesis of p-CA in the CEN.PK high-producing strain. This study demonstrates the importance of transporters in the engineering of cell factories for production of small molecules.

Evolutionary engineering reveals divergent paths when yeast is adapted to different acidic environments.

Tolerance of yeast to acid stress is important for many industrial processes including organic acid production. Therefore, elucidating the molecular basis of long term adaptation to acidic environments will be beneficial for engineering production strains to thrive under such harsh conditions. Previous studies using gene expression analysis have suggested that both organic and inorganic acids display similar responses during short term exposure to acidic conditions. However, biological mechanisms that will lead to long term adaptation of yeast to acidic conditions remains unknown and whether these mechanisms will be similar for tolerance to both organic and inorganic acids is yet to be explored. We therefore evolved Saccharomyces cerevisiae to acquire tolerance to HCl (inorganic acid) and to 0.3M L-lactic acid (organic acid) at pH 2.8 and then isolated several low pH tolerant strains. Whole genome sequencing and RNA-seq analysis of the evolved strains revealed different sets of genome alterations suggesting a divergence in adaptation to these two acids. An altered sterol composition and impaired iron uptake contributed to HCl tolerance whereas the formation of a multicellular morphology and rapid lactate degradation was crucial for tolerance to high concentrations of lactic acid. Our findings highlight the contribution of both the selection pressure and nature of the acid as a driver for directing the evolutionary path towards tolerance to low pH. The choice of carbon source was also an important factor in the evolutionary process since cells evolved on two different carbon sources (raffinose and glucose) generated a different set of mutations in response to the presence of lactic acid. Therefore, different strategies are required for a rational design of low pH tolerant strains depending on the acid of interest.

Enhanced amino acid utilization sustains growth of cells lacking Snf1/AMPK.

The metabolism of proliferating cells shows common features even in evolutionary distant organisms such as mammals and yeasts, for example the requirement for anabolic processes under tight control of signaling pathways. Analysis of the rewiring of metabolism, which occurs following the dysregulation of signaling pathways, provides new knowledge about the mechanisms underlying cell proliferation. The key energy regulator in yeast Snf1 and its mammalian ortholog AMPK have earlier been shown to have similar functions at glucose limited conditions and here we show that they also have analogies when grown with glucose excess. We show that loss of Snf1 in cells growing in 2% glucose induces an extensive transcriptional reprogramming, enhances glycolytic activity, fatty acid accumulation and reliance on amino acid utilization for growth. Strikingly, we demonstrate that Snf1/AMPK-deficient cells remodel their metabolism fueling mitochondria and show glucose and amino acids addiction, a typical hallmark of cancer cells.

Physiological characterization of the high malic acid-producing Aspergillus oryzae strain 2103a-68.

Malic acid is a C4 dicarboxylic acid that is currently mainly used in the food and beverages industry as an acidulant. Because of the versatility of the group of C4 dicarboxylic acids, the chemical industry has a growing interest in this chemical compound. As malic acid will be considered as a bulk chemical, microbial production requires organisms that sustain high rates, yields, and titers. Aspergillus oryzae is mainly known as an industrial enzyme producer, but it was also shown that it has a very competitive natural production capacity for malic acid. Recently, an engineered A. oryzae strain, 2103a-68, was presented which overexpressed pyruvate carboxylase, malate dehydrogenase, and a malic acid transporter. In this work, we report a detailed characterization of this strain including detailed rates and yields under malic acid production conditions. Furthermore, transcript levels of the genes of interest and corresponding enzyme activities were measured. On glucose as carbon source, 2103a-68 was able to secrete malic acid at a maximum specific production rate during stationary phase of 1.87 mmol (g dry weight (DW))⁻¹ h⁻¹ and with a yield of 1.49 mol mol⁻¹. Intracellular fluxes were obtained using ¹³C flux analysis during exponential growth, supporting the success of the metabolic engineering strategy of increasing flux through the reductive cytosolic tricarboxylic acid (rTCA) branch. Additional cultivations using xylose and a glucose/xylose mixture demonstrated that A. oryzae is able to efficiently metabolize pentoses and hexoses to produce malic acid at high titers, rates, and yields.

Data processing solutions to render metabolomics more quantitative: case studies in food and clinical metabolomics using Metabox 2.0.

In classic semiquantitative metabolomics, metabolite intensities are affected by biological factors and other unwanted variations. A systematic evaluation of the data processing methods is crucial to identify adequate processing procedures for a given experimental setup. Current comparative studies are mostly focused on peak area data but not on absolute concentrations. In this study, we evaluated data processing methods to produce outputs that were most similar to the corresponding absolute quantified data. We examined the data distribution characteristics, fold difference patterns between 2 metabolites, and sample variance. We used 2 metabolomic datasets from a retail milk study and a lupus nephritis cohort as test cases. When studying the impact of data normalization, transformation, scaling, and combinations of these methods, we found that the cross-contribution compensating multiple standard normalization (ccmn) method, followed by square root data transformation, was most appropriate for a well-controlled study such as the milk study dataset. Regarding the lupus nephritis cohort study, only ccmn normalization could slightly improve the data quality of the noisy cohort. Since the assessment accounted for the resemblance between processed data and the corresponding absolute quantified data, our results denote a helpful guideline for processing metabolomic datasets within a similar context (food and clinical metabolomics). Finally, we introduce Metabox 2.0, which enables thorough analysis of metabolomic data, including data processing, biomarker analysis, integrative analysis, and data interpretation. It was successfully used to process and analyze the data in this study. An online web version is available at http://metsysbio.com/metabox.

Discovery of procyanidin condensed tannins of (-)-epicatechin from Kratom, Mitragyna speciosa, as virucidal agents against SARS-CoV-2.

Kratom, Mitragyna speciosa, is one of the most popular herbs in the West and Southeast Asia. A number of previous works have focused on bioactive alkaloids in this plant; however, non-alkaloids have never been investigated for their biological activities. Antiviral and virucidal assays of a methanol leaf extract of Kratom, M. speciosa, revealed that a crude extract displayed virucidal activity against the SARS-CoV-2. Activity-guided isolation of a methanol leaf extract of Kratom led to the identification of B-type procyanidin condensed tannins of (-)-epicatechin as virucidal compounds against SARS-CoV-2. The fraction containing condensed tannins exhibited virucidal activity with an EC50 value of 8.38 µg/mL and a selectivity index (SI) value >23.86. LC-MS/MS analysis and MALDI-TOF MS identified the structure of the virucidal compounds in Kratom as B-type procyanidin condensed tannins, while gel permeation chromatograph (GPC) revealed weight average molecular weight of 238,946 Da for high molecular-weight condensed tannins. In addition to alkaloids, (-)-epicatechin was found as a major component in the leaves of M. speciosa, but it did not have virucidal activity. Macromolecules of (-)-epicatechin, i.e., procyanidin condensed tannins, showed potent virucidal activity against SARS-CoV-2, suggesting that the high molecular weights of these polyphenols are important for virucidal activity.

Quantifying fecal and plasma short-chain fatty acids in healthy Thai individuals.

Short-chain fatty acids (SCFAs) are involved in important physiological processes such as gut health and immune response, and changes in SCFA levels can be indicative of disease. Despite the importance of SCFAs in human health and disease, reference values for fecal and plasma SCFA concentrations in healthy individuals are scarce. To address this gap in current knowledge, we developed a simple and reliable derivatization-free GC-TOFMS method for quantifying fecal and plasma SCFAs in healthy individuals. We targeted six linear- and seven branched-SCFAs, obtaining method recoveries of 73-88% and 83-134% in fecal and plasma matrices, respectively. The developed methods are simpler, faster, and more sensitive than previously published methods and are well suited for large-scale studies. Analysis of samples from 157 medically confirmed healthy individuals showed that the total SCFAs in the feces and plasma were 34.1 ± 15.3 µmol/g and 60.0 ± 45.9 µM, respectively. In fecal samples, acetic acid (Ace), propionic acid (Pro), and butanoic acid (But) were all significant, collectively accounting for 89% of the total SCFAs, whereas the only major SCFA in plasma samples was Ace, constituting of 93% of the total plasma SCFAs. There were no statistically significant differences in the total fecal and plasma SCFA concentrations between sexes or among age groups. The data revealed, however, a positive correlation for several nutrients, such as carbohydrate, fat, iron from vegetables, and water, to most of the targeted SCFAs. This is the first large-scale study to report SCFA reference intervals in the plasma and feces of healthy individuals, and thereby delivers valuable data for microbiome, metabolomics, and biomarker research.

Measurement of very low-molecular weight metabolites by traveling wave ion mobility and its use in human urine samples.

The collision cross-sections (CCS) measurement using ion mobility spectrometry (IMS) in combination with mass spectrometry (MS) offers a great opportunity to increase confidence in metabolite identification. However, owing to the lack of sensitivity and resolution, IMS has an analytical challenge in studying the CCS values of very low-molecular-weight metabolites (VLMs ≤ 250 Da). Here, we describe an analytical method using ultrahigh-performance liquid chromatography (UPLC) coupled to a traveling wave ion mobility-quadrupole-time-of-flight mass spectrometer optimized for the measurement of VLMs in human urine samples. The experimental CCS values, along with mass spectral properties, were reported for the 174 metabolites. The experimental data included the mass-to-charge ratio (m/z), retention time (RT), tandem MS (MS/MS) spectra, and CCS values. Among the studied metabolites, 263 traveling wave ion mobility spectrometry (TWIMS)-derived CCS values (TWCCSN2) were reported for the first time, and more than 70% of these were CCS values of VLMs. The TWCCSN2 values were highly repeatable, with inter-day variations of <1% relative standard deviation (RSD). The developed method revealed excellent TWCCSN2 accuracy with a CCS difference (ΔCCS) within ±2% of the reported drift tube IMS (DTIMS) and TWIMS CCS values. The complexity of the urine matrix did not affect the precision of the method, as evidenced by ΔCCS within ±1.92%. According to the Metabolomics Standards Initiative, 55 urinary metabolites were identified with a confidence level of 1. Among these 55 metabolites, 53 (96%) were VLMs. The larger number of confirmed compounds found in this study was a result of the addition of TWCCSN2 values, which clearly increased metabolite identification confidence.

Rapid quantification of yeast lipid using microwave-assisted total lipid extraction and HPLC-CAD.

We here present simple and rapid methods for fast screening of yeast lipids in Saccharomyces cerevisiae. First we introduced a microwave-assisted technique for fast lipid extraction that allows the extraction of lipids within 10 min. The new method enhances extraction rate by 27 times, while maintaining product yields comparable to conventional methods (n = 14, P > 0.05). The recovery (n = 3) from spiking of synthetic standards were 92 ± 6% for cholesterol, 95 ± 4% for triacylglycerol, and 92 ± 4% for free fatty acids. Additionally, the new extraction method combines cell disruption and extraction in one step, and the approach, therefore, not only greatly simplifies sample handling but also reduces analysis time and minimizes sample loss during sample preparation. Second, we developed a chromatographic separation that allowed separation of neutral and polar lipids from the extracted samples within a single run. The separation was performed based on a three gradient solvent system combined with hydrophilic interaction liquid chromatography-HPLC followed by detection using a charged aerosol detector. The method was shown to be highly reproducible in terms of retention time of the analytes (intraday; 0.002-0.034% RSD; n = 10, interday; 0.04-1.35% RSD; n = 5) and peak area (intraday; 0.63-6% RSD; n = 10, interday; 4-12% RSD; n = 5).

Mass spectrometry-based analysis of gut microbial metabolites of aromatic amino acids.

Small molecules derived from gut microbiota have been increasingly investigated to better understand the functional roles of the human gut microbiome. Microbial metabolites of aromatic amino acids (AAA) have been linked to many diseases, such as metabolic disorders, chronic kidney diseases, inflammatory bowel disease, diabetes, and cancer. Important microbial AAA metabolites are often discovered via global metabolite profiling of biological specimens collected from humans or animal models. Subsequent metabolite identity confirmation and absolute quantification using targeted analysis enable comparisons across different studies, which can lead to the establishment of threshold concentrations of potential metabolite biomarkers. Owing to their excellent selectivity and sensitivity, hyphenated mass spectrometry (MS) techniques are often employed to identify and quantify AAA metabolites in various biological matrices. Here, we summarize the developments over the past five years in MS-based methodology for analyzing gut microbiota-derived AAA. Sample preparation, method validation, analytical performance, and statistical methods for correlation analysis are discussed, along with future perspectives.

Deep learning facilitates multi-data type analysis and predictive biomarker discovery in cancer precision medicine.

Cancer progression is linked to gene-environment interactions that alter cellular homeostasis. The use of biomarkers as early indicators of disease manifestation and progression can substantially improve diagnosis and treatment. Large omics datasets generated by high-throughput profiling technologies, such as microarrays, RNA sequencing, whole-genome shotgun sequencing, nuclear magnetic resonance, and mass spectrometry, have enabled data-driven biomarker discoveries. The identification of differentially expressed traits as molecular markers has traditionally relied on statistical techniques that are often limited to linear parametric modeling. The heterogeneity, epigenetic changes, and high degree of polymorphism observed in oncogenes demand biomarker-assisted personalized medication schemes. Deep learning (DL), a major subunit of machine learning (ML), has been increasingly utilized in recent years to investigate various diseases. The combination of ML/DL approaches for performance optimization across multi-omics datasets produces robust ensemble-learning prediction models, which are becoming useful in precision medicine. This review focuses on the recent development of ML/DL methods to provide integrative solutions in discovering cancer-related biomarkers, and their utilization in precision medicine.