Use of random forest in FTIR analysis of LDL cholesterol and tri‐glycerides for hyperlipidemia

A quantitative determination method for the diagnosis of hyperlipidemia was developed using Fourier transform infrared (FTIR) spectroscopy. Random forest (RF) was demonstrated as a potential multivariate algorithm for the FTIR analysis of low‐density lipoprotein cholesterol (LDL‐C) and tri‐glycerides (TG) in human serum samples. The informative wavebands for LDL‐C and TG were selected based on the Gini importance. The selected wavebands were mainly within the fingerprint region. The RF modeling results were better than those derived using PLS in validation process, because the chance for over‐fitting was possibly eliminated in RF algorithm. ARF also demonstrated favorable results in the test process. The prospective model exhibited a higher than 90% true prediction in negative/positive properties for male and female samples. These clinical statistical results indicated the optimization of RF algorithm performed accurately in the FTIR determination of LDL‐C and TG. RF is evaluated as a promising tool for diagnosing and controlling hyperlipidemia in populations. The parameter optimization methodology is useful in the improving model accuracy using FTIR spectroscopic technology. © 2015 American Institute of Chemical Engineers Biotechnol. Prog., 31:1693–1702, 2015     

Ouabain-induced modifications of prostate cancer cell lipidome investigated with mass spectrometry and FTIR spectroscopy

Fourier transform infrared (FTIR) spectroscopy was used to investigate modifications of prostate cancer PC-3 cell lipidome after exposure to sub-lethal concentrations of ouabain. FTIR spectroscopy offered an overview of the lipid classes present in the whole sample. The method is simple, label free and some features can be detected on entire cells. We compared the achievements of FTIR spectroscopy with data obtained by mass spectrometry (MS) on the same samples. It appears that FTIR spectroscopy could identify content variations in some lipid classes, e.g., these containing choline head groups such as phosphatidylcholine and sphingomyelin. MS analysis could confirm this result as indicated by principal component analysis and 2D heterocorrelation maps. FTIR spectra were also able to report changes in ester/choline/phosphate ratios characterizing lipid changes induced by ouabain. Furthermore, quantization of major lipid classes (PC, PE, PG, SM) could be obtained by curve fitting of the FTIR spectra. Yet, FTIR failed to resolve lipid classes for which the polar heads do not display specific IR features such as phosphatidylglycerol and cardiolipin.     

Application of multivariate curve resolution-alternating least squares (MCR-ALS) for secondary structure resolving of proteins

Fourier transform infrared (FTIR) spectroscopy is the most common spectroscopic technique used for study of protein structure. Initially, band deconvolution techniques were applied to determine the secondary structure of proteins. Recently, several multivariate regression methods have been used to predict the secondary structure of proteins as an alternative to the previous methods. Multivariate curve resolution-alternating least squares (MCR-ALS) was applied on the FTIR spectra of proteins to resolve the fraction and spectral profiles of different structural motifs. Initial estimates of spectral profiles of different protein motifs were built using orthogonal projection approach (OPA). Predicted fractions of α-helix and β-sheet obtained by MCR-ALS technique were compared with those from partial least squares (PLS) modeling which revealed superiority of the former. If we consider the possibility of pure spectra prediction in addition to the prediction of secondary structure from the data set, MCR-ALS can be proposed as a very valuable alternative for qualitative and quantitative study of protein structures.     

C-Reactive Protein, Waist Circumference, and Family History of Heart Attack are Independent Predictors of Body Iron Stores in Apparently Healthy Premenopausal Women

Ferritin levels have been associated with metabolic syndrome and insulin resistance. The aim of the present study was to evaluate the prediction of ferritin levels by variables related to cardiometabolic disease risk in a multivariate analysis. For this aim, 123 healthy women (72 premenopausal and 51 posmenopausal) were recruited. Data were collected through procedures of anthropometric measurements, questionnaires for personal/familial antecedents, and dietary intake (24-h recall), and biochemical determinations (ferritin, C reactive protein (CRP), glucose, insulin, and lipid profile) in blood serum samples obtained. Multiple linear regression analysis was used and variables with no normal distribution were log-transformed for this analysis. In premenopausal women, a model to explain log-ferritin levels was found with log-CRP levels, heart attack familial history, and waist circumference as independent predictors. Ferritin behaves as other cardiovascular markers in terms of prediction of its levels by documented predictors of cardiometabolic disease and related disorders. This is the first report of a relationship between heart attack familial history and ferritin levels. Further research is required to evaluate the mechanism to explain the relationship of central body fat and heart attack familial history with body iron stores values.     

Towards developing a protein infrared spectra databank (PISD) for proteomics research

Fourier transform infrared (FTIR) spectroscopy is an attractive tool for proteomics research as it can be used to rapidly characterize protein secondary structure in aqueous solution. Recently, a number of secondary structure prediction methods based on reference sets of FTIR spectra from proteins with known structure from X-ray crystallography have been suggested. These prediction methods, often referred to as pattern recognition based approaches, demonstrated good prediction accuracy using some error measure, e.g., the standard error of prediction (SEP). However, to avoid possible adverse effects from differences in recording, the analysis has been mostly based on reference sets of FTIR spectra from proteins recorded in one laboratory only. As a result, these studies were based on reference sets of FTIR spectra from a limited number of proteins. Pattern recognition based approaches, however, rely on reference sets of FTIR spectra from as many proteins as possible representing all possible band shape variation to be related to the diversity of protein structural classes. Hence, if we want to build reliable pattern recognition based systems to support proteomics research, which are capable of making good predictions from spectral data of any unknown protein, one common goal should be to build a comprehensive protein infrared spectra databank (PISD) containing FTIR spectra of proteins of known structure. We have started the process of developing a comprehensive PISD composed of spectra recorded in different laboratories. As part of this work, here we investigate possible effects on prediction accuracy achieved by a neural network analysis when using reference sets composed of FTIR spectra from different laboratories. Surprisingly low magnitude of difference in SEPs throughout all our experiments suggests that FTIR spectra recorded in different laboratories may be safely combined into one reference set with only minor deterioration of prediction accuracy in the worst case.     

Growth response of African catfish, Clarias gariepinus (Burchell) to varied protein and lipid levels

To study the growth response of African catfish (Clarias gariepinus) to the diets containing two protein (30 and 35%) and two lipid levels (12 and 18%) in a closed recirculatory system for 60 days, fish were fed to satiation twice daily. The experiment was carried out in triplicate groups. Fortnightly sampling of fish and water samples was carried out. The best result in terms of growth was obtained with the diet containing 34.62% protein and 12.12% lipid. The protein to lipid ratio was 18.57 in the diet that gave best percentage weight gain. However best survival, food conversion ratio and protein efficiency ratio were recorded in the diet containing 29.86% protein and 18.08% lipid.     

Cross-Species Extrapolation of Prediction Model for Lead Transfer from Soil to Corn Grain under Stress of Exogenous Lead

There has been increasing concern in recent years regarding lead (Pb) transfer in the soil-plant system. In this study the transfer of Pb (exogenous salts) was investigated from a wide range of Chinese soils to corn grain (Zhengdan 958). Prediction models were developed with combination of the Pb bioconcentration factor (BCF) of Zhengdan 958, and soil pH, organic matter (OM) content, and cation exchange capacity (CEC) through multiple stepwise regressions. Moreover, these prediction models from Zhengdan 958 were applied to other non-model corn species through cross-species extrapolation approach. The results showed that the soil pH and OM were the major factors that controlled Pb transfer from soil to corn grain. The lower pH and OM could improve the bioaccumulation of Pb in corn grain. No significant differences were found between two prediction models derived from the different exogenous Pb contents. When the prediction models were applied to other non-model corn species, the ratio ranges between the predicted BCF values and the measured BCF values were within an interval of 2-fold and close to the solid line of 1∶1 relationship. Moreover, the prediction model i.e. Log[BCF] = −0.098 pH-0.150 log[OM] −1.894 at the treatment of high Pb can effectively reduce the measured BCF intra-species variability for all non-model corn species. These suggested that this prediction model derived from the high Pb content was more adaptable to be applied to other non-model corn species to predict the Pb bioconcentration in corn grain and assess the ecological risk of Pb in different agricultural soils.     

Determination of solid-state fungal growth by Fourier transform infrared-photoacoustic spectroscopy

Abstract Photoacoustic spectroscopy (PAS) does not require optically transparent samples and is, therefore, well suited for analysis of solid-state samples. Fourier transform infrared (FTIR)-PAS of solid materials containing protein exhibited strong absorption in the amide I and amide II regions of the IR spectrum. Growth of a filamentous fungus, Phanerochaete chrysosporium , on cellulose discs was quantitatively determined by monitoring amide I absorption with FTIR-PAS.     

Cross-Species Extrapolation of Prediction Models for Cadmium Transfer from Soil to Corn Grain

Cadmium (Cd) is a highly toxic heavy metal for both plants and animals. The presence of Cd in agricultural soils is of great concern regarding its transfer in the soil-plant system. This study investigated the transfer of Cd (exogenous salts) from a wide range of Chinese soils to corn grain (Zhengdan 958). Through multiple stepwise regressions, prediction models were developed, with the combination of Cd bioconcentration factor (BCF) of Zhengdan 958 and soil pH, organic matter (OM) content, and cation exchange capacity (CEC). Moreover, these prediction models from Zhengdan 958 were applied to other non-model corn species through cross-species extrapolation approach. The results showed that the pH of the soil was the most important factor that controlled Cd uptake and lower pH was more favorable for Cd bioaccumulation in corn grain. There was no significant difference among three prediction models in the different Cd levels. When the prediction models were applied to other non-model corn species, the ratio ranges between the predicted BCF values and the measured BCF values were within an interval of 2 folds and close to the solid line of 1∶1 relationship. Furthermore, these prediction models also reduced the measured BCF intra-species variability for all non-model corn species. Therefore, the prediction models established in this study can be applied to other non-model corn species and be useful for predicting the Cd bioconcentration in corn grain and assessing the ecological risk of Cd in different soils.     

A Copula Based Approach for Design of Multivariate Random Forests for Drug Sensitivity Prediction

Modeling sensitivity to drugs based on genetic characterizations is a significant challenge in the area of systems medicine. Ensemble based approaches such as Random Forests have been shown to perform well in both individual sensitivity prediction studies and team science based prediction challenges. However, Random Forests generate a deterministic predictive model for each drug based on the genetic characterization of the cell lines and ignores the relationship between different drug sensitivities during model generation. This application motivates the need for generation of multivariate ensemble learning techniques that can increase prediction accuracy and improve variable importance ranking by incorporating the relationships between different output responses. In this article, we propose a novel cost criterion that captures the dissimilarity in the output response structure between the training data and node samples as the difference in the two empirical copulas. We illustrate that copulas are suitable for capturing the multivariate structure of output responses independent of the marginal distributions and the copula based multivariate random forest framework can provide higher accuracy prediction and improved variable selection. The proposed framework has been validated on genomics of drug sensitivity for cancer and cancer cell line encyclopedia database.     

Influence of physical and chemical properties of HTSXT-FTIR samples on the quality of prediction models developed to determine absolute concentrations of total proteins,carbohydrates and triglycerides: a preliminary study on the determination of their absolute concentrations in fresh microalgal biomass

Absolute concentrations of total macromolecules (triglycerides, proteins and carbohydrates) in microorganisms can be rapidly measured by FTIR spectroscopy, but caution is needed to avoid non-specific experimental bias. Here, we assess the limits within which this approach can be used on model solutions of macromolecules of interest. We used the Bruker HTSXT-FTIR system. Our results show that the solid deposits obtained after the sampling procedure present physical and chemical properties that influence the quality of the absolute concentration prediction models (univariate and multivariate). The accuracy of the models was degraded by a factor of 2 or 3 outside the recommended concentration interval of 0.5–35 µg spot^?1. Change occurred notably in the sample hydrogen bond network, which could, however, be controlled using an internal probe (pseudohalide anion). We also demonstrate that for aqueous solutions, accurate prediction of total carbohydrate quantities (in glucose equivalent) could not be made unless a constant amount of protein was added to the model solution (BSA). The results of the prediction model for more complex solutions, here with two components: glucose and BSA, were very encouraging, suggesting that this FTIR approach could be used as a rapid quantification method for mixtures of molecules of interest, provided the limits of use of the HTSXT-FTIR method are precisely known and respected. This last finding opens the way to direct quantification of total molecules of interest in more complex matrices.     

Cry1F protein not detected in soil after three years of transgenic Bt corn (1507 corn) use

To evaluate the potential of Bacillus thuringiensis (Bt) Cry1F protein accumulation in soil, transgenic corn containing event DAS-01507-1 encoding the cry1F gene was grown in three field sites for 3 consecutive yr, and the corn plants were incorporated into the soil through postseason tillage or no tillage each year. Soil samples were collected from these fields, and the level of Cry1F protein in these samples was determined using an enzyme-linked immunosorbent assay (ELISA) with a synthetic invertebrate gut fluid as an extraction buffer. The ELISA was validated in soil matrices over the concentration range of 18-180 ng/g dry weight, with a limit of detection of 4.5 ng/g dry weight. The assay was shown to have good accuracy and precision. No detectable Cry1F protein was found in any of the soil samples collected from the Cry1F corn fields. Soil also was bioassayed, and no biological activity was observed against Heliothis virescens neonates. These results indicate that the level of Cry1F protein accumulated in soil after 3-yr continuous planting of transgenic Cry1F corn is negligible.     

Interactions of proteins and cholesterol with lipids in bilayer membranes.

Mixtures of lipids and protein, the ATPase from rabbit sarcoplasmic reticulum, were studied by freeze-fracture electron microscopy and by measurement of the amount of fluid lipid with the spin label 2,2,6,6-tetramethylpiperidine-1-oxyl (TEM-PO). In dimyristoyl phosphatidylcholine vesicles the protein molecules were randomly distributed above the transition temperature, Tt, of the lipid and aggregated below Tt. For mixtures of dimyristoyl and dipalmitoyl phosphatidylcholine the existence of fluid and solid domains were shown in the temperature interval predicted from earlier TEMPO measurements. When protein was incorporated into this lipid mixture, freeze-fracture particles were randomly distributed in fluid lipids, or aggregated when only solid lipids were present. In mixtures of dimyristoyl phosphatidylcholine with cholesterol the protein was distributed randomly above the transition temperature of the phosphatidylcholine. Below that transition temperature the protein was excluded from a banded phase of solid lipid in the case of 10 mol% cholesterol. In mixtures containing 20 mol% cholesterol, protein molecules formed linear arrays, 50-200 nm in length, around smooth patches of lipid. Phase diagrams for lipid/cholesterol and lipid/protein systems are proposed which account for many of the available data. A model for increasing solidification of lipid around protein molecules or cholesterol above the transition temperature of the lipid is discussed.     

Characterisation of Authentic Lignin Biorefinery Samples by Fourier Transform Infrared Spectroscopy and Determination of the Chemical Formula for Lignin

Efficient methods for lignin characterisation are increasingly important as the field of lignin valorisation is growing with the increasing use of lignocellulosic feedstocks, such as wheat straw and corn stover, in biorefineries. In this study, we characterised a set of authentic lignin biorefinery samples in situ with no prior purification and minimal sample preparation. Lignin chemical formulas and lignin Fourier transform infrared (FTIR) spectra were extracted from mixed spectra by filtering out signals from residual carbohydrates and minerals. From estimations of C, H and O and adjustment for cellulose and hemicelluloses contents, the average chemical formula of lignin was found to be C₉H_10.2O_3.4 with slight variations depending on the biomass feedstock and processing conditions (between C₉H_9.5O_2.8 and C₉H_11.1O_3.6). Extracted FTIR lignin spectra showed many of the same characteristic peaks as organosolv and kraft lignin used as benchmark samples. Some variations in the lignin spectra of biorefinery lignin residue samples were found depending on biomass feedstock (wheat straw, corn stover or poplar) and on pretreatment severity, especially in the absorbance of bands at 1267 and 1032 cm^?1 relative to the strong band at ~1120 cm^?1. The suggested method of FTIR spectral analysis with adjustment for cellulose and hemicellulose is proposed to provide a fast and efficient way of analysing lignin in genuine lignin samples resulting from biorefineries.     

Alteration in Metabolic Signature and Lipid Metabolism in Patients with Angina Pectoris and Myocardial Infarction

Lipid metabolites are indispensable regulators of physiological and pathological processes, including atherosclerosis and coronary artery disease (CAD). However, the complex changes in lipid metabolites and metabolism that occur in patients with these conditions are incompletely understood. We performed lipid profiling to identify alterations in lipid metabolism in patients with angina and myocardial infarction (MI). Global lipid profiling was applied to serum samples from patients with CAD (angina and MI) and age-, sex-, and body mass index-matched healthy subjects using ultra-performance liquid chromatography/quadruple time-of-flight mass spectrometry and multivariate statistical analysis. A multivariate analysis showed a clear separation between the patients with CAD and normal controls. Lysophosphatidylcholine (lysoPC) and lysophosphatidylethanolamine (lysoPE) species containing unsaturated fatty acids and free fatty acids were associated with an increased risk of CAD, whereas species of lysoPC and lyso-alkyl PC containing saturated fatty acids were associated with a decreased risk. Additionally, PC species containing palmitic acid, diacylglycerol, sphingomyelin, and ceramide were associated with an increased risk of MI, whereas PE-plasmalogen and phosphatidylinositol species were associated with a decreased risk. In MI patients, we found strong positive correlation between lipid metabolites related to the sphingolipid pathway, sphingomyelin, and ceramide and acute inflammatory markers (high-sensitivity C-reactive protein). The results of this study demonstrate altered signatures in lipid metabolism in patients with angina or MI. Lipidomic profiling could provide the information to identity the specific lipid metabolites under the presence of disturbed metabolic pathways in patients with CAD.     

Applications of ATR-FTIR spectroscopic imaging to biomedical samples

FTIR spectroscopic imaging in ATR (Attenuated Total Reflection) mode is a powerful tool for studying biomedical samples. This paper summarises recent advances in the applications of ATR-FTIR imaging to dissolution of pharmaceutical formulations and drug release. The use of two different ATR accessories to obtain chemical images of formulations in contact with water as a function of time is demonstrated. The innovative use of the diamond ATR accessory allowed in situ imaging of tablet compaction and dissolution. ATR-FTIR imaging was also applied to obtain images of the surface of skin and the spatial distribution of protein and lipid rich domains was obtained. Chemical images of cross-section of rabbit aorta were obtained using a diamond ATR accessory and the possibility of in situ imaging of arterial samples in contact with aqueous solution was demonstrated for the first time. This experiment opens an opportunity to image arterial samples in contact with solutions containing drug molecules. This approach may help in understanding the mechanisms of treatment of atherosclerosis.     

Applications of ATR-FTIR spectroscopic imaging to biomedical samples

FTIR spectroscopic imaging in ATR (Attenuated Total Reflection) mode is a powerful tool for studying biomedical samples. This paper summarises recent advances in the applications of ATR-FTIR imaging to dissolution of pharmaceutical formulations and drug release. The use of two different ATR accessories to obtain chemical images of formulations in contact with water as a function of time is demonstrated. The innovative use of the diamond ATR accessory allowed in situ imaging of tablet compaction and dissolution. ATR-FTIR imaging was also applied to obtain images of the surface of skin and the spatial distribution of protein and lipid rich domains was obtained. Chemical images of cross-section of rabbit aorta were obtained using a diamond ATR accessory and the possibility of in situ imaging of arterial samples in contact with aqueous solution was demonstrated for the first time. This experiment opens an opportunity to image arterial samples in contact with solutions containing drug molecules. This approach may help in understanding the mechanisms of treatment of atherosclerosis.     

Growth and fatty acid composition of freshwater prawn, Macrobrachium rosenbergii, larvae fed diets containing various ratios of cod liver oil–corn oil mixture

A feeding trial was conducted to determine the effect of replacing costly cod liver oil with corn oil as a source of dietary lipid on the growth and fatty acid composition of the larval freshwater prawn, Macrobrachium rosenbergii de Man. Prawn larvae were weaned to artificial diets containing cod liver oil and corn oil either singly or in various combinations (2 : 1, 1 : 1, 1 : 2, w/w). Weaning to artificial diets from Artemia nauplii commenced at larval stage III with complete substitution by stage X. The reference group was reared solely on Artemia nauplii during the entire experiment. Incorporation of corn oil at 33–67% of dietary supplemental oil did not have significant effects on the post‐larval production. However, larvae fed with corn oil alone revealed a significantly lower post‐larval production compared to other experimental diets as well as to the reference group. No significant differences (P > 0.05) were observed in dry weight, protein and lipid concentration among larvae fed on various dietary treatments. Palmitic (16 : 0) and oleic/vaccenic (18 : 1) acids were the dominant saturated and monounsaturated fatty acids in larval tissues, respectively, whereas the polyunsaturated fraction was dominated by eicosapentaenoic (20 : 5n‐3) acid. The polyunsaturated fatty acid composition was dominated by n‐3 acids rather than n‐6 fatty acids. The fatty acid composition of the prawn in general reflected that of the diet. Larvae on diets containing higher concentrations of corn oil rich in linoleic (18 : 2n‐6) acid showed a higher concentration of this acid in their tissues. No evidence of de novo synthesis of linoleic (18 : 2n‐6) acid was found. Higher levels of stearic (18 : 0), arachidonic (20 : 4n‐6) and eicosapentaenoic (20 : 5n‐3) acids found in larvae as compared with those fed Artemia and artificial diets strongly indicated the larval ability in chain elongation and desaturation of palmitic (16 : 0), linoleic (18 : 2n‐6) or linolenic (18 : 3n‐3) acids, respectively. Despite a large variation of n‐3 to n‐6 ratios of the live and artificial diets, larval n‐3 to n‐6 ratios were relatively stable among different dietary treatments, possibly indicative of the importance of such a ratio in the larval fatty acid metabolism.     

Hepatic uptake of chylomicrons and triglyceride emulsions in rats fed diets of differing fat content

The hepatic removal of plasma chylomicrons was determined for rats fed the following diets: a) containing no triglyceride, b) regular chow diet with 4.5% of its mass as lipid and, c) a corn oil-supplemented chow with triglyceride accounting for 20% of the mass. The fractional hepatic uptake of either radiolabeled chylomicrons or a triglyceride emulsion was reciprocally related to the amount of lipid in the diet. The animals receiving only carbohydrate and protein calories had the most active hepatic uptake of particulate triglyceride and were observed to have a significant decrease in the plasma concentration of the C apolipoproteins. The addition of either C-I, C-II, or C-III apoproteins to the triglyceride emulsion prior to intravenous injection produced a significantly lower hepatic triglyceride recovery of emulsions containing apoC-III. When the plasma of animals fed a fat-free diet was supplemented with human C-III-1 apolipoprotein, the distribution into the liver of either enterally administered fatty acid or parenteral triglyceride was diminished. The triglyceride content in the liver of the rats fed fat-free or corn oil-supplemented diets was significantly greater than that of the control rats and composition was somewhat similar to that of lymph triglyceride. The studies indicate an important influence of dietary lipid on both the partition of plasma triglyceride into the liver and the steady state hepatic triglyceride content.     

Feasibility of Spectroscopic Characterization of Algal Lipids: Chemometric Correlation of NIR and FTIR Spectra with Exogenous Lipids in Algal Biomass

A large number of algal biofuels projects rely on a lipid screening technique for selecting a particular algal strain with which to work. We have developed a multivariate calibration model for predicting the levels of spiked neutral and polar lipids in microalgae, based on infrared (both near-infrared (NIR) and Fourier transform infrared (FTIR)) spectroscopy. The advantage of an infrared spectroscopic technique over traditional chemical methods is the direct, fast, and non-destructive nature of the screening method. This calibration model provides a fast and high-throughput method for determining lipid content, providing an alternative to laborious traditional wet chemical methods. We present data of a study based on nine levels of exogenous lipid spikes (between 1% and 3% (w/w)) of trilaurin as a triglyceride and phosphatidylcholine as a phospholipid model compound in lyophilized algal biomass. We used a chemometric approach to corrrelate the main spectral changes upon increasing phospholipid and triglyceride content in algal biomass collected from single species. A multivariate partial least squares (PLS) calibration model was built and improved upon with the addition of multiple species to the dataset. Our results show that NIR and FTIR spectra of biomass from four species can be used to accurately predict the levels of exogenously added lipids. It appears that the cross-species verification of the predictions is more accurate with the NIR models (R ²?=?0.969 and 0.951 and RMECV?=?0.182 and 0.227% for trilaurin and phosphatidylcholine spike respectively), compared with FTIR (R ²?=?0.907 and 0.464 and RMECV?=?0.302 and 0.767% for trilaurin and phosphatidylcholine spike, respectively). A fast high-throughput spectroscopic lipid fingerprinting method can be applied in a multitude of screening efforts that are ongoing in the microalgal research community.