No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Understanding the information encoded in the human genome and the influence of each part of the DNA sequence is a fundamental problem of our society that can be key to unveil the mechanism of common diseases. With the latest technological developments in the genomics field, many research institutes have the tools to collect massive amounts of genomic data. Nevertheless, there is a lack of tools that can be used to process and analyse these datasets in a biologically reliable and efficient manner. Many deep learning solutions have been proposed to solve current genomic tasks, but most of the times the main research interest is in the underlying biological mechanisms rather than high scores of the predictive metrics themselves. Recently, state-of-the-art in deep learning has shifted towards large transformer models, which use an attention mechanism that can be leveraged for interpretability. The main drawbacks of these large models is that they require a lot of memory space and have high inference time, which may make their use unfeasible in practical applications. In this work, we test the appropriateness of knowledge distillation to obtain more usable and equally performing models that genomic researchers can easily fine-tune to solve their scientific problems. DNABERT, a transformer model pre-trained on DNA data, is distilled following two strategies: DistilBERT and MiniLM. Four student models with different sizes are obtained and fine-tuned for promoter identification. They are evaluated in three key aspects: classification performance, usability and biological relevance of the predictions. The latter is assessed by visually inspecting the attention maps of TATA-promoter predictions, which are expected to have a peak of attention at the well-known TATA motif present in these sequences. Results show that is indeed possible to obtain significantly smaller models that are equally performant in the promoter identification task without any major differences between the two techniques tested. The smallest distilled model experiences less than 1% decrease in all performance metrics evaluated (accuracy, F1 score and Matthews Correlation Coefficient) and an increase in the inference speed by 7.3x, while only having 15% of the parameters of DNABERT. The attention maps for the student models show that they successfully learn to mimic the general understanding of the DNA that DNABERT possesses.

Dihydrodiol dehydrogenases are a family of aldo-keto reductases (AKR1Cs) involved in the metabolism of steroid hormones and xenobiotics. Whilst, several phase II drugs as well as endogenous & exogenous steroids/steroid metabolites have been identified as inducers of gene transcription, the cellular transcription factors controlling the expression of AKR1C1 are incompletely elucidated. Herein, we have cloned and characterized the proximal promoter region of the human AKR1C1 gene that controls its transcription. The 5’ flanking proximal promoter region of the AKR1C1 gene consists of a TATA box and an inverted CCAAT binding site. Deletion analysis of the 5’-flanking, ~3.0 kb region of the human AKR1C1 gene identified the region between -128 to -88 as the minimal proximal promoter essential for basal transcription of AKR1C1 in human ovarian (2008 & 2008/C13*), lung (H23 & A549) and liver carcinoma (HepG2) cells. Antioxidant response elements (ARE) have been shown to modulate the transcription ofv genes coding for phase II drug metabolizing enzymes. Cloning of the ARE upstream of the AKR1C1 proximal promoter resulted in increased transcription in human lung adenocarcinoma and liver hepatoblastoma cells but not in human ovarian carcinoma cells. Further, ARE increased the induction of the AKR1C1 gene in response to treatment with phase II drug inducers. However, ARE did not induce the transcription of AKR1C1 gene promoter in the presence of cisplatin in any of the cell lines. A computational analysis utilizing the Alibaba 2.0 on the proximal AKR1C1 gene promoter region was performed to identify potential transcription factor binding sites. Based on this analysis, a set of potential, putative transcription factor binding sites for Oct1, Sp1, Cp-1/NF-Y, CEBP, p40X, USF, NF1 and AP-2 were identified in the region -180 to -88 of the AKR1C1 gene promoter. Site-directed mutagenesis studies indicated that the transcription factor binding sites for NF-Y/CEBP were involved in controlling the basal transcription of AKR1C1 in all the cancer cells studied. Electrophoretic mobility shift (EMSAs) and gel supershift assays demonstrated that the transcription factor NF-Y preferentially binds to the inverted CCAAT box at -109ATTGG-105 of the AKR1C1 gene. Chromatin immunoprecipitation (ChIP) analysis confirmed the in vivo association between NF-Y and human AKR1C1 gene promoter in human ovarian, lung and liver carcinoma cells. Further, ectopic expression of NF-Y’s increased the AKR1C1 gene transcription, whereas expression of a dominant-negative NF-YA or knockdown of NF-YA by siRNA transfection, decreased the AKR1C1 gene transcription. A 2-fold increase in AKR1C1 transcription was observed specifically in cisplatin-treated 2008 cells that was CCAAT box-dependent. These results indicate that NF-Y regulates basal transcription of AKR1C1 in human ovarian, lung and liver carcinoma cells and cisplatin-induced transcription in human ovarian carcinoma cells. / Pathology

Herpes simplex virus (HSV) genes are expressed as at least three coordinately regulated gene classes during lytic infection. The delayed-early (DE) and late (L) genes require previous expression of one or more immediate-early (IE) genes for their own expression. The DE genes achieve maximal expression prior to viral DNA synthesis, while the L genes are maximally expressed after DNA replication (Honess and Roizman, 1974). A recombinant strain of HSV-1, X1N17, was used in this study to examine the effect of the gene promoter on the temporal expression of HSV genes. This virus carries a late viral promoter upstream from the coding sequences of a DE gene (thymidine kinase; TK). S1-mapping studies showed that X1N17-TK transcripts initiated under the control of the late promoter and accumulated with L class kinetics. However, the TK activity levels in X1N17-infected cells were not consistent with HSV late gene expression. Western blot analysis of infected cell proteins revealed that despite the high levels of X1N17-TK mRNA present in the cytoplasm late after infection, little TK polypeptide was being synthesized. This suggested that HSV genes are subject to post-transcriptional control mechanisms that modulate the efficiency of translation of viral transcripts. More specifically, it appears as though HSV-TK transcripts are not efficiently translated at late times in infected cells. / Thesis / Master of Science (MS)

The cyst nematode, Globodera tabacum tabacum Behrens, and the parasitic angiosperm, Egyptian broomrape, Orobanche aegyptiaca Pers., are obligate root parasites that cause severe yield and quality loss of many important crop hosts. Although these represent two diverse classes of parasites, they have significant similarities in the modes of parasitism and complex interactions with their hosts. Conventional control methods have had limited success in controlling these parasites. The overall objective of this research was to engineer resistance to the cyst nematode and Egyptian broomrape by expressing genes encoding parasite specific toxins under the control of parasite-responsive promoters using tobacco (Nicotiana tabacum L. cv. Xanthi). For nematode resistance, an anti-feeding strategy was employed utilizing the tomato proteinase inhibitor I (PI-I) gene as a nematode specific toxin. Transgenic tobacco plants were generated that expressed genes encoding an intracellarly retained or secreted form of tomato PI-I under the control of the nematode-inducible promoter, derived from tomato (Lycopersicon esculentum L.) Hmg2 gene. Our goals were to determine the effectiveness of local PI-I expression on nematode resistance and to determine if intracellular or extracellular PI-I deposition enhances resistance. Two constructs were generated that contained either the coding region of the tomato PI-I gene, lacking the signal sequence (EM1), or the coding region of PI-I including the signal sequence (EM2), fused to the nematode-responsive Hmg2 promoter. Transgenic PI-I plants were inoculated with G. t. tabacum cysts and evaluated for nematode interactions. Our results suggest that local expression of intercellular of PI-I significantly reduced cyst production when compared to the nontransformed controls. For broomrape resistance, a well characterized R/avr gene pair, the tobacco N resistance gene and the tobacco mosaic virus replicase (TMV) gene, was utilized to create novel gene-for-gene resistance via a N gene-mediated hypersensitive response (HR) to limit broomrape parasitism. The bean (Phaselous vulgaris L.) chalcone synthase 8 (CHS8) promoter has been characterized as a broomrape–responsive promoter. We introduced the CHS8:TMV replicase gene construct into tobacco plants that contains an endogenous N gene. Transgenic tobacco plants were inoculated with O. aegyptiaca seeds and monitored for parasite attachment and development. The expression of the TMV replicase leads to a significant reduction in broomrape parasitism. These genetic engineering strategies show promise in enhancing resistance to these destructive parasites. / Ph. D.