Towards spontaneous facial expression recognition in real-world video

Facial expression is an important channel of human social communication. Facial expression recognition (FER) aims to perceive and understand emotional states of humans based on information in the face. Building robust and high performance FER systems that can work in real-world video is still a challenging task, due to the various unpredictable facial variations and complicated exterior environmental conditions, as well as the difficulty of choosing a suitable type of feature descriptor for extracting discriminative facial information. Facial variations caused by factors such as pose, age, gender, race and occlusion, can exert profound influence on the robustness, while a suitable feature descriptor largely determines the performance. Most present attention on FER has been paid to addressing variations in pose and illumination. No approach has been reported on handling face localization errors and relatively few on overcoming facial occlusions, although the significant impact of these two variations on the performance has been proved and highlighted in many previous studies. Many texture and geometric features have been previously proposed for FER. However, few comparison studies have been conducted to explore the performance differences between different features and examine the performance improvement arisen from fusion of texture and geometry, especially on data with spontaneous emotions. The majority of existing approaches are evaluated on databases with posed or induced facial expressions collected in laboratory environments, whereas little attention has been paid on recognizing naturalistic facial expressions on real-world data. This thesis investigates techniques for building robust and high performance FER systems based on a number of established feature sets. It comprises of contributions towards three main objectives: (1) Robustness to face localization errors and facial occlusions. An approach is proposed to handle face localization errors and facial occlusions using Gabor based templates. Template extraction algorithms are designed to collect a pool of local template features and template matching is then performed to covert these templates into distances, which are robust to localization errors and occlusions. (2) Improvement of performance through feature comparison, selection and fusion. A comparative framework is presented to compare the performance between different features and different feature selection algorithms, and examine the performance improvement arising from fusion of texture and geometry. The framework is evaluated for both discrete and dimensional expression recognition on spontaneous data. (3) Evaluation of performance in the context of real-world applications. A system is selected and applied into discriminating posed versus spontaneous expressions and recognizing naturalistic facial expressions. A database is collected from real-world recordings and is used to explore feature differences between standard database images and real-world images, as well as between real-world images and real-world video frames. The performance evaluations are based on the JAFFE, CK, Feedtum, NVIE, Semaine and self-collected QUT databases. The results demonstrate high robustness of the proposed approach to the simulated localization errors and occlusions. Texture and geometry have different contributions to the performance of discrete and dimensional expression recognition, as well as posed versus spontaneous emotion discrimination. These investigations provide useful insights into enhancing robustness and achieving high performance of FER systems, and putting them into real-world applications.

Design and construction of an in-plant activation cassette for transgene expression and recombinant protein production in plants

Virus-based transgene expression systems have become particularly valuable for recombinant protein production in plants. The dual-module in-plant activation (INPACT) expression platform consists of a uniquely designed split-gene cassette incorporating the cis replication elements of Tobacco yellow dwarf geminivirus (TYDV) and an ethanol-inducible activation cassette encoding the TYDV Rep and RepA replication-associated proteins. The INPACT system is essentially tailored for recombinant protein production in stably transformed plants and provides both inducible and high-level transient transgene expression with the potential to be adapted to diverse crop species. The construction of a novel split-gene cassette, the inducible nature of the system and the ability to amplify transgene expression via rolling-circle replication differentiates this system from other DNA- and RNA-based virus vector systems used for stable or transient recombinant protein production in plants. Here we provide a detailed protocol describing the design and construction of a split-gene INPACT cassette, and we highlight factors that may influence optimal activation and amplification of gene expression in transgenic plants. By using Nicotiana tabacum, the protocol takes 6-9 months to complete, and recombinant proteins expressed using INPACT can accumulate to up to 10% of the leaf total soluble protein.

Random Gabor based templates for facial expression recognition in images with facial occlusion

Robust facial expression recognition (FER) under occluded face conditions is challenging. It requires robust algorithms of feature extraction and investigations into the effects of different types of occlusion on the recognition performance to gain insight. Previous FER studies in this area have been limited. They have spanned recovery strategies for loss of local texture information and testing limited to only a few types of occlusion and predominantly a matched train-test strategy. This paper proposes a robust approach that employs a Monte Carlo algorithm to extract a set of Gabor based part-face templates from gallery images and converts these templates into template match distance features. The resulting feature vectors are robust to occlusion because occluded parts are covered by some but not all of the random templates. The method is evaluated using facial images with occluded regions around the eyes and the mouth, randomly placed occlusion patches of different sizes, and near-realistic occlusion of eyes with clear and solid glasses. Both matched and mis-matched train and test strategies are adopted to analyze the effects of such occlusion. Overall recognition performance and the performance for each facial expression are investigated. Experimental results on the Cohn-Kanade and JAFFE databases demonstrate the high robustness and fast processing speed of our approach, and provide useful insight into the effects of occlusion on FER. The results on the parameter sensitivity demonstrate a certain level of robustness of the approach to changes in the orientation and scale of Gabor filters, the size of templates, and occlusions ratios. Performance comparisons with previous approaches show that the proposed method is more robust to occlusion with lower reductions in accuracy from occlusion of eyes or mouth.

Expression of a bacterial xylanase in Trichoderma reesei under the egl2 and cbh2 glycosyl hydrolase gene promoters

Expression vectors were constructed for Trichoderma reesei using the promoters, secretion signals and the modular structure of the efficiently expressed and secreted cellulase enzymes EGL2 (Cel5A) and CBH2 (Cel6A) as a prelude to establishing a platform where a gene of interest can be expressed under several promoters simultaneously. The designs featured (i) EGL2sigpro (egl2 promoter and secretion signal), (ii) EGL2cbmlin (egl2 promoter, secretion signal, EGL2 cellulose binding module and linker), (iii) CBH2sigpro (cbh2 promoter and secretion signal) and (iv) CBH2cbmlin (cbh2 promoter, secretion signal, CBH2 cellulose binding module and linker). Recombinant vectors were introduced individually into the high protein-secreting T. reesei RUT-C30 strain to generate single-promoter transformants expressing the Dictyoglomus thermophilum xynB gene that encodes a thermophilic xylanase enzyme (XynB). Ten transformants producing XynB representing each of the four different types of vectors were selected for further testing and the highest XynB production was achieved from a transformant containing 1–2 copies of the EGL2cbmlin vector. Best xylanase producers did not show any particular pattern in terms of the number of gene copies and their mode of integration into the chromosomal DNA. Transformants generated with the cbmlin-type vectors produced multiple forms of XynB which were decorated with various N- and O-glycans. One of the O-glycans was identified as hexuronic acid, whose presence had not been observed previously in the glycosylation patterns of T. reesei.

Transient expression vectors for functional genomics, quantification of promoter activity and RNA silencing in plants

Background We describe novel plasmid vectors for transient gene expression using Agrobacterium, infiltrated into Nicotiana benthamiana leaves. We have generated a series of pGreenII cloning vectors that are ideally suited to transient gene expression, by removing elements of conventional binary vectors necessary for stable transformation such as transformation selection genes. Results We give an example of expression of heme-thiolate P450 to demonstrate effectiveness of this system. We have also designed vectors that take advantage of a dual luciferase assay system to analyse promoter sequences or post-transcriptional regulation of gene expression. We have demonstrated their utility by co-expression of putative transcription factors and the promoter sequence of potential target genes and show how orthologous promoter sequences respond to these genes. Finally, we have constructed a vector that has allowed us to investigate design features of hairpin constructs related to their ability to initiate RNA silencing, and have used these tools to study cis-regulatory effect of intron-containing gene constructs. Conclusion In developing a series of vectors ideally suited to transient expression analysis we have provided a resource that further advances the application of this technology. These minimal vectors are ideally suited to conventional cloning methods and we have used them to demonstrate their flexibility to investigate enzyme activity, transcription regulation and post-transcriptional regulatory processes in transient assays.

Enhancement of bacterial mutagenicity of bifunctional alkylating agents by expression of mammalian glutathione s-transferase

Recently, we inserted the plasmid vector pKK233-2 containing rat GSH S-transferase (GST) 5-5 cDNA into Salmonella typhimurium TA1535 and found that these bacteria [GST 5-5(+)] expressed the protein and produced mutations when ethylene or methylene dihalides were added [Thier, R., Taylor, J. B., Pemble, S. E., Ketterer, B., Persmark, M., Humphreys, W. G., and Guengerich, F. P. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 8576-8580]. After exposure to the known GST 5-5 substrate 1,2-epoxy-3-(4′-nitrophenoxy)propane, the GST 5-5(+) strain showed fewer mutants than the bacteria transfected with the cDNA clone in a reverse orientation [GST 5-5(-)], suggesting a protective role of GST 5-5. However, mutations were considerably enhanced in the GST 5-5(+) strain [as compared to GST 5-5(-)] when 1,2,3,4-diepoxybutane (butadiene diepoxide) or 1,2-epoxy-4-bromobutane was added. The GST 5-5(+) and GST 5-5(-) bacterial stains showed similar responses to 1,2-epoxypropane, 3,4-epoxy-1-butene, and 1,4-dibromobutane. The results suggest that some bifunctional activated butanes are transformed to mutagenic products through GSH conjugation. We also found that the GST 5-5(+) strain showed enhanced mutagenicity with 1,4-dibromo-2,3-epoxybutane, 1,2-epoxy-3-bromopropane (epibromohydrin), and (±)-1,4-dibromo-2,3-dihydroxybutane. The possibility was considered that a 5-membered thialonium ion may be involved in the mutagenicity. Model thialonium compounds were rather stable to hydrolysis in aqueous solution at pH 7.4 and slowly alkylated 4-(4-nitrobenzyl)pyridine. The presence of a hydroxyl group β to the sulfur did not enhance reactivity. Mechanisms involving episulfonium ions are considered more likely. Potential oxidation products of the toxic pesticide 1,2-dibromo-3-chloropropane (DBCP) were also considered in this system. DBCP itself gave rather similar results in the two strains. Others have reported that oxidation of DBCP is required for mutagenicity, along with GST-catalyzed GSH conjugation [Simula, T. P., Glancey, M. J., Söderlund, E. J., Dybing, E., and Wolf, C. R. (1993) Carcinogenesis 14, 2303-2307]. The putative oxidation product 1,2-dibromopropional did not show a difference between the two strains. However, 1,3-dichloroacetone, a model for the putative oxidation product 1-bromo-3-chloroacetone, was considerably more mutagenic in the GST 5-5(+) strain.

Expression of mammalian glutathione S-transferase 5-5 in Salmonella typhimurium TA1535 leads to base-pair mutations upon exposure to dihalomethanes

Dihalomethanes can produce liver tumors in mice but not in rats, and concern exists about the risk of these compounds to humans. Glutathione (GSH) conjugation of dihalomethanes has been considered to be a critical event in the bioactivation process, and risk assessment is based upon this premise; however, there is little experimental support for this view or information about the basis of genotoxicity. A plasmid vector containing rat GSH S-transferase 5-5 was transfected into the Salmonella typhimurium tester strain TA1535, which then produced active enzyme. The transfected bacteria produced base-pair revertants in the presence of ethylene dihalides or dihalomethanes, in the order CH2Br2 > CH2BrCl > CH2Cl2. However, revertants were not seen when cells were exposed to GSH, CH2Br2, and an amount of purified GSH S-transferase 5-5 (20-fold excess in amount of that expressed within the cells). HCHO, which is an end product of the reaction of GSH with dihalomethanes, also did not produce mutations. S-(1-Acetoxymethyl)GSH was prepared as an analog of the putative S-(1-halomethyl)GSH reactive intermediates. This analog did not produce revertants, consistent with the view that activation of dihalomethanes must occur within the bacteria to cause genetic damage, presenting a model to be considered in studies with mammalian cells. S-(1-Acetoxymethyl)GSH reacted with 2′-deoxyguanosine to yield a major adduct, identified as S-[1-(N2-deoxyguanosinyl)methyl]GSH. Demonstration of the activation of dihalomethanes by this mammalian GSH S-transferase theta class enzyme should be of use in evaluating the risk of these chemicals, particularly in light of reports of the polymorphic expression of a similar activity in humans.

Expression, Purification and Characterization of Peanut (Arachis hypogaea) Agglutinin (PNA) from Baculovirus Infected Insect Cells

Peanut (Arachis hypogaea) seed lectin, PNA is widely used to identify tumor specific antigen (T-antigen), Gal beta 1-3GalNAc on the eukaryotic cell surface. The functional amino acid coding region of a cDNA clone, pBSH-PN was PCR amplified and cloned downstream of the polyhedrin promoter in the Autographa californica nucleopolyhedrovirus (AcNPV) based transfer vector pVL1393. Co-transfection of Spodoptera frugiperda cells (Sf9) with the transfer vector, pAcPNA and AcRP6 (a recombinant AcNPV having B-gal downstream of the polyhedrin promoter) DNAs produced a recombinant virus, AcPNA which expresses PNA. Infection of suspension culture of Sf9 cells with plaque purified AcPNA produced as much as 9.8 mg PNA per liter (2.0 x 10(6) cells/ml) of serum-free medium. Intracellularly expressed protein (re-PNA) was purified to apparent homogeneity by affinity chromatography using ECD-Sepharose. Polyclonal antibodies against natural PNA (n-PNA) crossreacted with re-PNA. The subunit molecular weight (30 kDa), hemagglutination activity, and carbohydrate specificity of re-PNA were found to be identical to that of n-PNA, thus confirming the abundant production of a functionally active protein in the baculovirus expression system.

Analysis of a Genomic DNA Expression Library of Mycobacterium tuberculosis Using Tuberculosis Patient Sera: Evidence for Modulation of Host Immune Response

DNA obtained from a human sputum isolate of Mycobacterium tuberculosis, NTI-64719, which showed extensive dissemination in the guinea pig model resulting in a high score for virulence was used to construct an expression library in the lambda ZAP vector. The size of DNA inserts in the library ranged from 1 to 3 kb, and recombinants represented 60% of the total plaques obtained. When probed with pooled serum from chronically infected tuberculosis patients, the library yielded 176 recombinants with a range of signal intensities. Among these, 93 recombinants were classified into 12 groups on the basis of DNA hybridization experiments, The polypeptides synthesized by the recombinants were predominantly LacZ fusion proteins, Serum obtained from patients who were clinically diagnosed to be in the early phase of M. tuberculosis infection was used to probe the 176 recombinants obtained. interestingly, some recombinants that gave very strong signals in the original screen did not react with early-phase serum; conversely, others whose signals were extremely weak in the original screen gave very intense signals with serum from recently infected patients, This indicates the differential nature of either the expression of these antigens or the immune response elicited by them as a function of disease progression.

Comparative immunogenicity of M. hyopneumoniae NrdF encoded in different expression systems delivered orally via attenuated S. typhimurium aroA in mice.

The Mycoplasma hyopneumoniae ribonucleotide reductase R2 subunit (NrdF) gene fragment was cloned into eukaryotic and prokaryotic expression vectors and its immunogenicity evaluated in mice immunized orally with attenuated Salmonella typhimurium aroA CS332 harboring either of the recombinant expression plasmids. We found that NrdF is highly conserved among M. hyopneumoniae strains. The immunogenicity of NrdF was examined by analyzing antibody responses in sera and lung washes, and the cell-mediated immune (CMI) response was assessed by determining the INF-[gamma] level produced by splenocytes upon in vitro stimulation with NrdF antigen. S. typhimurium expressing NrdF encoded by the prokaryotic expression plasmid (pTrcNrdF) failed to elicit an NrdF-specific serum or secretory antibody response, and IFN-[gamma] was not produced. Similarly, S. typhimurium carrying the eukaryotic recombinant plasmid encoding NrdF (pcNrdF) did not induce a serum or secretory antibody response, but did elicit significant NrdF-specific IFN-[gamma] production, indicating induction of a CMI response. However, analysis of immune responses against the live vector S. typhimurium aroA CS332 showed a serum IgG response but no mucosal IgA response in spite of its efficient invasiveness in vitro. In the present study we show that the DNA vaccine encoding the M. hyopneumoniae antigen delivered orally via a live attenuated S. typhimurium aroA can induce a cell-mediated immune response. We also indicate that different live bacterial vaccine carriers may have an influence on the type of the immune response induced.

Pheo-type: A diagnostic gene-expression assay for the classification of pheochromocytoma and paraganglioma

Context: Pheochromocytomas and paragangliomas (PPGLs) are heritable neoplasms that can be classified into gene-expression subtypes corresponding to their underlying specific genetic drivers. Objective: This study aimed to develop a diagnostic and research tool (Pheo-type) capable of classifying PPGL tumors into gene-expression subtypes that could be used to guide and interpret genetic testing, determine surveillance programs, and aid in elucidation of PPGL biology. Design: A compendium of published microarray data representing 205 PPGL tumors was used for the selection of subtype-specific genes that were then translated to the Nanostring gene-expression platform. A support vector machine was trained on the microarray dataset and then tested on an independent Nanostring dataset representing 38 familial and sporadic cases of PPGL of known genotype (RET, NF1, TMEM127, MAX, HRAS, VHL, and SDHx). Different classifier models involving between three and six subtypes were compared for their discrimination potential. Results: A gene set of 46 genes and six endogenous controls was selected representing six known PPGL subtypes; RTK1–3 (RET, NF1, TMEM127, and HRAS), MAX-like, VHL, and SDHx. Of 38 test cases, 34 (90%) were correctly predicted to six subtypes based on the known genotype to gene-expression subtype association. Removal of the RTK2 subtype from training, characterized by an admixture of tumor and normal adrenal cortex, improved the classification accuracy (35/38). Consolidation of RTK and pseudohypoxic PPGL subtypes to four- and then three-class architectures improved the classification accuracy for clinical application. Conclusions: The Pheo-type gene-expression assay is a reliable method for predicting PPGL genotype using routine diagnostic tumor samples.

Prediction of Candidate Primary Immunodeficiency Disease Genes Using a Support Vector Machine Learning Approach

Screening and early identification of primary immunodeficiency disease (PID) genes is a major challenge for physicians. Many resources have catalogued molecular alterations in known PID genes along with their associated clinical and immunological phenotypes. However, these resources do not assist in identifying candidate PID genes. We have recently developed a platform designated Resource of Asian PDIs, which hosts information pertaining to molecular alterations, protein-protein interaction networks, mouse studies and microarray gene expression profiling of all known PID genes. Using this resource as a discovery tool, we describe the development of an algorithm for prediction of candidate PID genes. Using a support vector machine learning approach, we have predicted 1442 candidate PID genes using 69 binary features of 148 known PID genes and 3162 non-PID genes as a training data set. The power of this approach is illustrated by the fact that six of the predicted genes have recently been experimentally confirmed to be PID genes. The remaining genes in this predicted data set represent attractive candidates for testing in patients where the etiology cannot be ascribed to any of the known PID genes.

Part I: Enzyme replacement versus neurotrophic factor viral vector-mediated gene therapy of Parkinson’s disease, Part II: The effects of orally dosed tolcapone and entacapone on dopamine metabolism in the dorsal striatum and nucleus accumbens in rats

Part I: Parkinson’s disease is a slowly progressive neurodegenerative disorder in which particularly the dopaminergic neurons of the substantia nigra pars compacta degenerate and die. Current conventional treatment is based on restraining symptoms but it has no effect on the progression of the disease. Gene therapy research has focused on the possibility of restoring the lost brain function by at least two means: substitution of critical enzymes needed for the synthesis of dopamine and slowing down the progression of the disease by supporting the functions of the remaining nigral dopaminergic neurons by neurotrophic factors. The striatal levels of enzymes such as tyrosine hydroxylase, dopadecarboxylase and GTP-CH1 are decreased as the disease progresses. By replacing one or all of the enzymes, dopamine levels in the striatum may be restored to normal and behavioral impairments caused by the disease may be ameliorated especially in the later stages of the disease. The neurotrophic factors glial cell derived neurotrophic factor (GDNF) and neurturin have shown to protect and restore functions of dopaminergic cell somas and terminals as well as improve behavior in animal lesion models. This therapy may be best suited at the early stages of the disease when there are more dopaminergic neurons for neurotrophic factors to reach. Viral vector-mediated gene transfer provides a tool to deliver proteins with complex structures into specific brain locations and provides long-term protein over-expression. Part II: The aim of our study was to investigate the effects of two orally dosed COMT inhibitors entacapone (10 and 30 mg/kg) and tolcapone (10 and 30 mg/kg) with a subsequent administration of a peripheral dopadecarboxylase inhibitor carbidopa (30 mg/kg) and L- dopa (30 mg/kg) on dopamine and its metabolite levels in the dorsal striatum and nucleus accumbens of freely moving rats using dual-probe in vivo microdialysis. Earlier similarly designed studies have only been conducted in the dorsal striatum. We also confirmed the result of earlier ex vivo studies regarding the effects of intraperitoneally dosed tolcapone (30 mg/kg) and entacapone (30 mg/kg) on striatal and hepatic COMT activity. The results obtained from the dorsal striatum were generally in line with earlier studies, where tolcapone tended to increase dopamine and DOPAC levels and decrease HVA levels. Entacapone tended to keep striatal dopamine and HVA levels elevated longer than in controls and also tended to elevate the levels of DOPAC. Surprisingly in the nucleus accumbens, dopamine levels after either dose of entacapone or tolcapone were not elevated. Accumbal DOPAC levels, especially in the tolcapone 30 mg/kg group, were elevated nearly to the same extent as measured in the dorsal striatum. Entacapone 10 mg/kg elevated accumbal HVA levels more than the dose of 30 mg/kg and the effect was more pronounced in the nucleus accumbens than in the dorsal striatum. This suggests that entacapone 30 mg/kg has minor central effects. Also our ex vivo study results obtained from the dorsal striatum suggest that entacapone 30 mg/kg has minor and transient central effects, even though central HVA levels were not suppressed below those of the control group in either brain area in the microdialysis study. Both entacapone and tolcapone suppressed hepatic COMT activity more than striatal COMT activity. Tolcapone was more effective than entacapone in the dorsal striatum. The differences between dopamine and its metabolite levels in the dorsal striatum and nucleus accumbens may be due to different properties of the two brain areas.

Dendritic Poly(ether imine) Based Gene Delivery Vector

The nonviral vector based gene delivery approach is attractive due to advantages associated with molecular-level modifications suitable for optimization of vector properties. In a new class of nonviral gene delivery systems, we herein report the potential of poly(ether Mine) (PETIM) dendrimers to mediate an effective gene delivery function. PETIM dendrimer, constituted with tertiary amine branch points, n-propyl ether linkers and primary amines at their peripheries, exhibits significantly reduced toxicities, over a broad concentration range. The dendrimer complexes pDNA effectively, protects DNA from endosomal damages, and delivers to the cell nucleus. Gene transfection studies, utilizing a reporter plasmid pEGFP-C1 and upon complexation with dendrimer, showed a robust expression of the encoded protein. The study shows that PETIM dendrimers are hitherto unknown novel gene delivery vectors, combining features of poly(ethylene imine)-based polymers and dendrimers, yet are relatively nontoxic and structurally precise.

High level stable expression of rat brain type IIA sodium channel ?-subunit in CHO cells

The neuronal sodium channels are responsible for the rising phase of action potential and are composed of three subunits, of which the alpha-subunit has been shown to be adequate for most of its functional properties. We have stably expressed the rat brain type IIA sodium channel alpha-subunit in CHO cell tine using a CMV promoter-based vector. The expression was confirmed by detecting a 6.5 kb RNA corresponding to sodium channel alpha-subunit using Northern hybridization. The cells stably expressing the alpha-subunit, yield isolated sodium currents of amplitudes greater than 4nA when studied in whole-cell configuration of the patch-clamp technique. The sodium currents are characterized by activation and inactivation properties similar to neuronal sodium channels, and are blocked by the voltage gated sodium channel blocker tetrodotoxin (TTX).