Double-check: validation of diagnostic statistics for PLS-DA models in metabolomics studies

Partial Least Squares-Discriminant Analysis (PLS-DA) is a PLS regression method with a special binary ‘dummy’ y-variable and it is commonly used for classification purposes and biomarker selection in metabolomics studies. Several statistical approaches are currently in use to validate outcomes of PLS-DA analyses e.g. double cross validation procedures or permutation testing. However, there is a great inconsistency in the optimization and the assessment of performance of PLS-DA models due to many different diagnostic statistics currently employed in metabolomics data analyses. In this paper, properties of four diagnostic statistics of PLS-DA, namely the number of misclassifications (NMC), the Area Under the Receiver Operating Characteristic (AUROC), Q ² and Discriminant Q ² (DQ ²) are discussed. All four diagnostic statistics are used in the optimization and the performance assessment of PLS-DA models of three different-size metabolomics data sets obtained with two different types of analytical platforms and with different levels of known differences between two groups: control and case groups. Statistical significance of obtained PLS-DA models was evaluated with permutation testing. PLS-DA models obtained with NMC and AUROC are more powerful in detecting very small differences between groups than models obtained with Q ² and Discriminant Q ² (DQ ²). Reproducibility of obtained PLS-DA models outcomes, models complexity and permutation test distributions are also investigated to explain this phenomenon. DQ ² and Q ² (in contrary to NMC and AUROC) prefer PLS-DA models with lower complexity and require higher number of permutation tests and submodels to accurately estimate statistical significance of the model performance. NMC and AUROC seem more efficient and more reliable diagnostic statistics and should be recommended in two group discrimination metabolomic studies.

     

Comparative analysis of creatinine and osmolality as urine normalization strategies in targeted metabolomics for the differential diagnosis of asthma and COPD

Introduction

Urine is an ideal matrix for metabolomics investigation due to its non-invasive nature of collection and its rich metabolite content. Despite the advancements in mass spectrometry and ¹H-NMR platforms in urine metabolomics, the statistical analysis of the generated data is challenged with the need to adjust for the hydration status of the person. Normalization to creatinine or osmolality values are the most adopted strategies, however, each technique has its challenges that can hinder its wider application. We have been developing targeted urine metabolomic methods to differentiate two important respiratory diseases, namely asthma and chronic obstructive pulmonary disease (COPD).

Objective

To assess whether the statistical model of separation of diseases using targeted metabolomic data would be improved by normalization to osmolality instead of creatinine.

Methods

The concentration of 32 metabolites was previously measured by two liquid chromatography-tandem mass spectrometry methods in 51 human urine samples with either asthma (n?=?25) or COPD (n?=?26). The data was normalized to creatinine or osmolality. Statistical analysis of the normalized values in each disease was performed using partial least square discriminant analysis (PLS-DA). Models of separation of diseases were compared.

Results

We found that normalization to creatinine or osmolality did not significantly change the PLS-DA models of separation (R²Q²?=?0.919, 0.705 vs R²Q²?=?0.929, 0.671, respectively). The metabolites of importance in the models remained similar for both normalization methods.

Conclusion

Our findings suggest that targeted urine metabolomic data can be normalized for hydration using creatinine or osmolality with no significant impact on the diagnostic accuracy of the model.

     

The invariance of production per unit of food consumed in fish populations

The amount of biomass production per unit of food consumed (P/Q) represents an important quantity in ecosystem functioning, because it indicates how efficient a population transforms ingested food into biomass. Several investigations have noticed that P/Q remains relatively constant (or invariant) across fish population that feed at the same food-type level (carnivorous/herbivorous). Nevertheless, theoretical explanation for this invariant is still lacking. In this paper, we demonstrate that P/Q remains invariant across fish populations with stable-age distribution. Three key assumptions underpin the P/Q invariant: (1) the ratio between natural mortality M and von Bertalanffy growth parameter k (M/k ratio) should remain invariant across fish populations; (2) a parameter defining the fraction of ingested food available for growth needs to remain constant across fish that feed at the same trophic level; (3) third, the ratio between length at age 0 (\(l_0\)) and asymptotic length (\(l_\infty\)) should be constant across fish populations. The influence of these assumptions on the P/Q estimates were numerically assessed considering fish populations of different lifespan. Numerical evaluations show that the most critical condition highly relates to the first assumption, M/k. Results are discussed in the context of the reliability of the required assumption to consider the P/Q invariant in stable-age distributed fish populations.     

Genetic analysis of the traits introgressed from <Emphasis Type="Italic">Aegilops speltoides</Emphasis> Tausch. to bread wheat and determined by chromosome 5A genes

A winter bread wheat accession from the Arsenal collection was genetically examined to study the results of introgression, which substantially changed the physiological and morphological traits of the original spring cultivar Rodina. Apart from its winter habit, the accession was characterized by awned speltoid spikes, suggesting introgression into chromosome 5A, which carries marker genes in the order Vrn-A1-Q-B1. Genetic analysis showed that the chromosome fragment introgressed from Aegilops speltoides recombined well with the homeologous region of bread wheat chromosome 5A in the region between the Vrn-A1 and Q genes. Recombination between the Vrn-A1 and B1 genes was not detected, and it was assumed that the order of the marker genes of chromosome 5A was inverted to produce Q-Vrn-A1-B1. When the winter introgression line was crossed with Triticum spelta L., an interaction of two dominant genes determining the spike character was for the first time detected in F₁, increasing the spike length and the number of spikelets, and followed with transgression in F₂. It was assumed that Ae. speltoides had a homeoallelic speltoid gene, which was designated as Q ^S .     

Evaluation of <Emphasis Type="Italic">Galleria mellonella</Emphasis> larvae as an in vivo model for assessing the relative toxicity of food preservative agents

Larvae of Galleria mellonella are widely used for evaluating the virulence of microbial pathogens and for measuring the efficacy of anti-microbial agents and produce results comparable to those that can be obtained using mammals. In this work, the suitability of using G. mellonella larvae to measure the relative toxicity of a variety of food preservatives was evaluated. The response of larvae to eight commonly used food preservatives (potassium nitrate, potassium nitrite, potassium sorbate, sodium benzoate, sodium nitrate, sodium chloride, sodium nitrite and sodium acetate) administered by feeding or by intra-haemocoel injection was measured. A significant correlation between the LD₅₀ (R ²?=?0.8766, p?=?0.0006) and LD₈₀ (R ²?=?0.7629, p?=?0.0046) values obtained due to oral or intra-haemocoel administration of compounds was established. The response of HEp-2 cells to the food preservatives was determined, and a significant correlation (R ²?=?0.7217, p?=?0.0076) between the LD₅₀ values of the compounds administered by feeding in larvae with the IC₅₀ values of the compounds in HEp-2 cells was established. A strong correlation between the LD₅₀ values of the eight food preservatives in G. mellonella larvae and rats (R ²?=?0.6506, p?=?0.0156) was demonstrated. The results presented here indicate that G. mellonella larvae may be used as a model to evaluate the relative toxicity of food preservatives, and the results show a strong positive correlation to those obtained using established cell culture and mammalian models.     

Dynamic changes in energy metabolism and electron transport of photosystem II in <Emphasis Type="Italic">Nicotiana tabacum</Emphasis> infested by nymphs of <Emphasis Type="Italic">Bemisia tabaci</Emphasis> (Middle East-Asia Minor 1)

Bemisia tabaci Middle East-Asia Minor 1 (MEAM1) infestation adversely affected photosynthesis of host plants. In the current study, chlorophyll a fluorescence was measured to determine the effects of MEAM1 nymph infestation of tobacco local and systemic leaves on energy metabolism and electron transport of photosystemII(PSII). The results showed that the density of PSII reaction centres per excited cross section (CS) (RC/CS) of infested and systemic leaves was reduced at 14 and 20 days. In systemic leaves, the number of PSII closed reaction centres (1-qP) increased significantly at 14 and 20 days. Absorption flux per Q_A^? reducing PSII reaction centre (RC) (ABS/RC), trapped energy flux per RC (TRo/RC), and electron transport per RC (ETo/RC) of infested and systemic leaves increased with MEAM1 nymph infestation. The most obvious increase in absorption flux per CS (ABS/CSo) and trapped energy flux per CS (TRo/CSo) of infested and systemic leaves occurred at 14 days. MEAM1 nymph infestation significantly reduced the energy required for PSII Q_A to be completely reduced (Sm) in tobacco leaves. These results suggested that MEAM1 nymph infestation caused changes in light-harvesting antenna system and deactivation of the reaction centre, resulting in the reduction of photons absorbed by reaction centres per unit area. MEAM1 nymph infestation, particularly the 3rd instar nymphs, decreased light utilization ability and increased excess excitation energy in tobacco leaves. With MEAM1 nymph infestation, the relative electron transport capacity of the entire electron transport chain decreased, and more light energy was used to reduce Q_A.     

Solution of Levinthal’s paradox is possible at the level of the formation and assembly of protein secondary structures

The complete volume of a protein’s conformation space is smaller by many orders of magnitude at the level of secondary-structure elements as compared with the conformation of amino-acid residues. According to Levinthal’s estimate, the latter is ~10^2L, with L being the number of residues in the chain, while the former, at the level of secondary structures, increases no faster than ~L^N with N being the number of the secondary-structure elements. N is approximately L/15 according to the statistics of protein structures. This drastic decrease in the exponent (L/15 instead of 2L) considerably reduces the sampling space and explains the reason that sampling of the conformation space at the level of secondary-structure elements does not prevent the protein chain from finding its most stable structure.     

Search for genes influencing the maintenance of the [<Emphasis Type="Italic">ISP</Emphasis><Superscript>+</Superscript>] prion-like antisuppressor determinant in yeast with the use of an insertion gene library

The prion-like determinant [ISP ⁺] manifests itself as an antisuppressor of certain sup35 mutations. To establish that [ISP ⁺] is indeed a new yeast prion, it is necessary to identify the gene that codes for the protein whose prion form is [ISP ⁺]. Analysis of the transformants obtained by transformation of an [ISP ⁺] strain with an insertion gene library revealed three genes controlling the [ISP ⁺] maintenance: UPF1, UPF2, and SFP1. SFP1 codes for a potentially prionogenic protein, which is enriched in Asn and Gln residues, and is thereby the most likely candidate for the [ISP ⁺] structural gene. UPF1 and UPF2 code for components of nonsense-mediated mRNA decay. The [ISP ⁺] elimination caused by UPF1 and UPF2 inactivation was reversible, and Upf1p and Upf2p were not functionally related to phosphatase Ppz1p, which influences the [ISP ⁺] manifestation. Possible mechanisms sustaining the influence of UPF1 and UPF2 on [ISP ⁺] maintenance are discussed.     

Fluorescence Induction Kinetics in Membrane Preparations of Photosystem II with Heterogeneous Metal Clusters (Mn/Fe) in the Oxygen-Evolving Complex

Transport of electrons in spinach photosystem II (PSII) whose oxygen-evolving complex (OEC) contains heterogeneous metal clusters 2Mn2Fe and 3Mn1Fe was studied by measuring the fluorescence induction kinetics (FIK). PSII(2Mn,2Fe) and PSII(3Mn,1Fe) preparations were produced using Cadepleted PSII membranes (PSII(–Ca)). It was found that FIK in PSII(2Mn,2Fe) membranes is similar in form to FIK in PSII(–Ca) samples, but the fluorescence yield is lower in PSII(2Mn,2Fe). The results demonstrate that, just as in PSII(–Ca) preparations, there is electron transfer from the metal cluster in the OEC to the primary plastoquinone electron acceptor Q_A. They also show that partial substitution of Mn cations with Fe has no effect on the electron transport on the acceptor side of PSII. Thus, these data demonstrate the possibility of water oxidation either by the heterogeneous metal cluster or just by the manganese dimer. We established that FIK in PSII(3Mn,1Fe) preparations are similar in form to FIK in PSII(2Mn,2Fe) membranes but PSII(3Mn,1Fe) is characterized by a slightly higher maximal fluorescence yield, F_max. The electron transfer rate in PSII(3Mn,1Fe) preparations significantly (by a factor of two) increases in the presence of Ca²⁺, whereas Ca²⁺ has hardly any effect on the electron transport in PSII(2Mn,2Fe) membranes. In Mndepleted PSII membranes, FIK reaches its maximum (the so-called peak K), after which the fluorescence yield starts to decrease as the result of two factors: the oxidation of reduced primary plastoquinone Q _A ^? and the absence of electron influx from the donor side of PSII. The replacement of Mn cations by Fe in PSII(?Mn) preparations leads to fluorescence saturation and disappearance of the K peak. This is possibly due to the deceleration of the charge recombination process that takes place between reduced primary electron acceptor Q _A ^? and oxidized tyrosine Y _Z ^+. which is an electron carrier between the OEC and the primary electron donor P680.     

Bio-optical characterization of selected cyanobacteria strains present in marine and freshwater ecosystems

The optical properties, i.e., absorption and scattering spectra of ten strains of cyanobacteria from the Baltic Sea and Pomeranian lakes (Aphanizomenon flos-aquae KAC 15, Microcystis aeruginosa CCNP 1101, Anabaena sp. CCNP 1406, Synechocystis salina CCNP 1104, Phormidium sp. CCNP 1317, Nodularia spumigena CCNP 1401, Synechococcus sp. CCNP 1108, Nostoc sp. CCNP 1411, Cyanobacterium sp. CCNP 1105, Pseudanabaena cf. galeata CCNP 1312) grown under low light conditions were investigated. Moreover, the chlorophylls, carotenoids, and phycobilin composition as well as the size structure of chosen cyanobacteria were measured. Studied species revealed high diversity both in optical properties with the absorption spectra similarity index ranging from 0.67 to 0.94 and the pigment composition. The chlorophyll-specific absorption coefficient at 440 nm a _ph *(440) varied between 0.017 and 0.065 m² mg^?1. The influence of the package effect was only observed in the case of large filamentous cyanobacteria like N. spumigena or Nostoc sp. Interestingly, the package effect factor Q _a *(675) for large-celled Anabaena sp. was 0.92. Besides chlorophyll a, only echinenone, β-carotene, and phycocyanin were present in all analyzed cyanobacteria strains. Zeaxanthin, which is widely used as a marker pigment for cyanobacteria, was absent in the toxic N. spumigena and Anabaena sp., which are the species that occur in the Baltic Sea most frequently causing summer cyanobacterial blooms. The investigation also showed that the sample preservation technique can introduce some major errors within the absorption band affected by the phycocyanin absorption.     

Exocyst Complex Member EXOC5 Is Required for Survival of Hair Cells and Spiral Ganglion Neurons and Maintenance of Hearing

The exocyst, an octameric protein complex consisting of Exoc1 through Exoc8, was first determined to regulate exocytosis by targeting vesicles to the plasma membrane in yeast to mice. In addition to this fundamental role, the exocyst complex has been implicated in other cellular processes. In this study, we investigated the role of the exocyst in cochlear development and hearing by targeting EXOC5, a central exocyst component. Deleting Exoc5 in the otic epithelium with widely used Cre lines resulted in early lethality. Thus, we generated two different inner ear-specific Exoc5 knockout models by crossing Gfi1^Cre mice with Exoc5^f/f mice for hair cell-specific deletion (Gfi1^Cre/+;Exoc5^f/f) and by in utero delivery of rAAV-iCre into the otocyst of embryonic day 12.5 for deletion throughout the otic epithelium (rAAV2/1-iCre;Exoc5^f/f). Gfi1^Cre/+;Exoc5^f/f mice showed relatively normal hair cell morphology until postnatal day 20, after which hair cells underwent apoptosis accompanied by disorganization of stereociliary bundles, resulting in progressive hearing loss. rAAV2/1-iCre;Exoc5^f/f mice exhibited abnormal neurite morphology, followed by apoptotic degeneration of spiral ganglion neurons (SGNs) and hair cells, which led to profound and early-onset hearing loss. These results demonstrate that Exoc5 is essential for the normal development and survival of cochlear hair cells and SGNs, as well as the functional maintenance of hearing.     

Karyotypic study of five Lutjanid species using conventional and Ag-NORs banding techniques

The first cytogenetic comparisons of five snapper species from Thailand were presented here. Renal cell samples were taken from blacktail snapper (Lutjanus fulvus), five lined snapper (L. quinquelineatus), dory snapper (L. fulviflamma), brownstripe red snapper (L. vitta), and mangrove red snapper (L. argentimaculatus). The mitotic chromosome preparation was prepared directly from kidney cells. Conventional staining and Ag-NOR banding techniques were applied to stain the chromosomes. The results exhibited that all five snapper species have the diploid chromosome numbers of 2n = 48 and the fundamental numbers (NF) of 48. The presences of large, medium, and small telocentric chromosomes were 22-24-2, 24-20-4, 36-10-2, 28-16-4 and 36-10-2, respectively. The Ag- NORs banding technique provides the pair of nucleolar organizer regions (NORs) at subcentromeric region of the long arm of the respective telocentric chromosome pairs 9, 1, 3, 4 and 9. Their karyotype formulas is as follows: L. fulvus (2n = 48): L ₂₂ ^t + M ₂₄ ^t + S ₂ ^t , L. quinquelineatus (2n = 48): L ₂₄ ^t + M ₂₀ ^t + S ₄ ^t , L. fulviflamma (2n = 48): Lt36 + Mt10 + St2, L. vitta (2n = 48): L ₂₈ ^t + M ₁₆ ^t + S ₄ ^t , and L. argentimaculatus (2n = 48): L ₃₆ ^t + M ₁₀ ^t + S ₂ ^t .     

Phosphorus intensity determines short-term P uptake by pigeon pea (<Emphasis Type="Italic">Cajanus cajan</Emphasis> L.) grown in soils with differing P buffering capacity

Phosphorus (P) uptake by plant roots depends on P intensity (I) and P quantity (Q) in the soil. The relative importance of Q and I on P uptake is unknown for soils with large P sorption capacities because of difficulties in determining trace levels of P in the soil solution. We applied a new isotope based method to detect low P concentrations (<20 μg P l⁻¹). The Q factor was determined by assessment of the isotopically exchangeable P in the soil (E-value) and the I factor was determined by measurement of the P concentration in the pore water. A pot trial was set up using four soils with similar labile P quantities but contrasting P buffering capacities. Soils were amended with KH₂PO₄ at various rates and pigeon pea (Cajanus cajan L.) was grown for 25 days. The P intensity ranged between 0.0008 and 50 mg P l⁻¹ and the P quantity ranged between 10 and 500 mg P kg⁻¹. Shoot dry matter (DM) yield and P uptake significantly increased with increasing P application rates in all soils. Shoot DM yield and P uptake, relative to the maximal yield or P uptake, were better correlated with the P concentration in the pore water (R ² = 0.83–0.90) than with the E-value (R ²=0.40–0.53). The observed P uptakes were strongly correlated to values simulated using a mechanistic rhizosphere model (NST 3.0). A sensitivity analysis reveals that the effect of P intensity on the short-term P uptake by pigeon pea exceeded the effect of P quantity both at low and high P levels. However, DM yield and P uptake at a given P intensity consistently increased with increasing P buffering capacity (PBC). The experimental data showed that the intensity yielding 80% of the maximal P uptake was 4 times larger in the soil with the smallest PBC compared to the soil with the largest PBC. This study confirms that short-term P uptake by legumes is principally controlled by the P intensity in the soil, but is to a large extent also affected by the PBC of the soil. Section Editor: N. J. Barrow     

On the minimum electron transport coefficients in tokamaks in the range of low collision frequencies

There are two close empirical scalings, namely, the T-11 and neo-Alcator ones, that provide correct estimates for the energy confinement time in tokamaks in ohmic heating regimes in the linear part of the dependence τ _E (\(\bar n_e \)) in the range of low values of \(\bar n_e \) and 〈ν _e ^* 〉 ≤ 1. The similar character of electron energy confinement in this range, which expands with increasing magnetic field B ₀, has stimulated the search for dimensionless parameters and simple physical models that would explain the experimentally observed dependences χ _e ～ 1/n _e and τ _Ee ～ \(\bar n_e \). In 1987, T. Okhawa showed that the experimental data were satisfactorily described by the formula χ_e⊥ = (c ²/ω _pe ² )ν _e /qR, in deriving of which the random spatial leap along the radius r on the electron trajectory was assumed to be the same as that in the coefficient of the poloidal field diffusion, while the repetition rate of these leaps was assumed to be ν _e /qR. In 2004, J. Callen took into account the decrease in the fraction of transient electrons with increasing toroidal ratio ? = r/R and corrected the coefficient c ²/ω _pe ² in Okhawa equation by the factor σ _‖ ^Sp /σ _‖ ^neo . If one takes into account this correction and assumes that the frequency of the stochastic process is equal to the reciprocal of the half-period of rotation of a trapped electron along its banana trajectory, then the resulting expression for χ_e⊥ will coincide with the T-11 scaling: χ _e ^an ∞ ?^1.75(T _e /A _i )^0.5/(n _e qR) at A i = 1. If the same stochastic process also involves ions, it may result in the opening of the orbit of a trapped ion at the distance ～(c/ω _pe )(m _i /m _e )^1/4. In this case, the calculated coefficient of electron and ion diffusion D is close to D ^an ≈ χ _e ^an /2.     

Oxygen transfer as a tool for fine-tuning recombinant protein production by <Emphasis Type="Italic">Pichia pastoris</Emphasis> under glyceraldehyde-3-phosphate dehydrogenase promoter

Effects of oxygen transfer on recombinant protein production by Pichia pastoris under glyceraldehyde-3-phosphate dehydrogenase promoter were investigated. Recombinant glucose isomerase was chosen as the model protein. Two groups of oxygen transfer strategies were applied, one of which was based on constant oxygen transfer rate where aeration rate was Q _O/V = 3 and 10 vvm, and agitation rate was N = 900 min^?1; while the other one was based on constant dissolved oxygen concentrations, C _DO = 5, 10, 15, 20 and 40 % in the fermentation broth, by using predetermined exponential glucose feeding with μ _o = 0.15 h^?1. The highest cell concentration was obtained as 44 g L^?1 at t = 9 h of the glucose fed-batch phase at C _DO = 20 % operation while the highest volumetric and specific enzyme activities were obtained as 4440 U L^?1 and 126 U g^?1 cell, respectively at C _DO = 15 % operation. Investigation of specific enzyme activities revealed that keeping C _DO at 15 % was more advantageous with an expense of relatively higher by-product formation and lower specific cell growth rate. For this strategy, the highest oxygen transfer coefficient and oxygen uptake rate were K _L a = 0.045 s^?1 and OUR = 8.91 mmol m^?3 s^?1, respectively.