Sentiments about Mental Health on Twitter-Before and during the COVID-19 Pandemic

During the COVID-19 pandemic, the novel coronavirus had an impact not only on public health but also on the mental health of the population. Public sentiment on mental health and depression is often captured only in small, survey-based studies, while work based on Twitter data often only looks at the period during the pandemic and does not make comparisons with the pre-pandemic situation. We collected tweets that included the hashtags #MentalHealth and #Depression from before and during the pandemic (8.5 months each). We used LDA (Latent Dirichlet Allocation) for topic modeling and LIWC, VADER, and NRC for sentiment analysis. We used three machine-learning classifiers to seek evidence regarding an automatically detectable change in tweets before vs. during the pandemic: (1) based on TF-IDF values, (2) based on the values from the sentiment libraries, (3) based on tweet content (deep-learning BERT classifier). Topic modeling revealed that Twitter users who explicitly used the hashtags #Depression and especially #MentalHealth did so to raise awareness. We observed an overall positive sentiment, and in tough times such as during the COVID-19 pandemic, tweets with #MentalHealth were often associated with gratitude. Among the three classification approaches, the BERT classifier showed the best performance, with an accuracy of 81% for #MentalHealth and 79% for #Depression. Although the data may have come from users familiar with mental health, these findings can help gauge public sentiment on the topic. The combination of (1) sentiment analysis, (2) topic modeling, and (3) tweet classification with machine learning proved useful in gaining comprehensive insight into public sentiment and could be applied to other data sources and topics.

Do the Math: Making Mathematics in Wikipedia Computable

Wikipedia combines the power of AI solutions and human reviewers to safeguard article quality. Quality control objectives include detecting malicious edits, fixing typos, and spotting inconsistent formatting. However, no automated quality control mechanisms currently exist for mathematical formulae. Spell checkers are widely used to highlight textual errors, yet no equivalent tool exists to detect algebraically incorrect formulae. Our paper addresses this shortcoming by making mathematical formulae computable. We present a method that (1) gathers the semantic information surrounding the context of each mathematical formulae, (2) provides access to the information in a graph-structured dependency hierarchy, and (3) performs automatic plausibility checks on equations. We evaluate the performance of our approach on 6,337 mathematical expressions contained in 104 Wikipedia articles on the topic of orthogonal polynomials and special functions. Our system, [Formula: see text], verified 358 out of 1,516 equations as error-free. [Formula: see text] successfully translated 27% of the mathematical expressions and outperformed existing translation approaches by 16%. Additionally, [Formula: see text] achieved an F1 score of .495 for annotating mathematical expressions with relevant textual descriptions, which is a significant step towards advancing searchability, readability, and accessibility of mathematical formulae in Wikipedia. A prototype of [Formula: see text] and the semantically enhanced Wikipedia articles are available at: https://tpami.wmflabs.org.

Caching and Reproducibility: Making Data Science Experiments Faster and FAIRer

Small to medium-scale data science experiments often rely on research software developed ad-hoc by individual scientists or small teams. Often there is no time to make the research software fast, reusable, and open access. The consequence is twofold. First, subsequent researchers must spend significant work hours building upon the proposed hypotheses or experimental framework. In the worst case, others cannot reproduce the experiment and reuse the findings for subsequent research. Second, suppose the ad-hoc research software fails during often long-running computational expensive experiments. In that case, the overall effort to iteratively improve the software and rerun the experiments creates significant time pressure on the researchers. We suggest making caching an integral part of the research software development process, even before the first line of code is written. This article outlines caching recommendations for developing research software in data science projects. Our recommendations provide a perspective to circumvent common problems such as propriety dependence, speed, etc. At the same time, caching contributes to the reproducibility of experiments in the open science workflow. Concerning the four guiding principles, i.e., Findability, Accessibility, Interoperability, and Reusability (FAIR), we foresee that including the proposed recommendation in a research software development will make the data related to that software FAIRer for both machines and humans. We exhibit the usefulness of some of the proposed recommendations on our recently completed research software project in mathematical information retrieval.

Category-Based Deep CCA for Fine-Grained Venue Discovery From Multimodal Data

In this work, travel destinations and business locations are taken as venues. Discovering a venue by a photograph is very important for visual context-aware applications. Unfortunately, few efforts paid attention to complicated real images such as venue photographs generated by users. Our goal is fine-grained venue discovery from heterogeneous social multimodal data. To this end, we propose a novel deep learning model, category-based deep canonical correlation analysis. Given a photograph as input, this model performs: 1) exact venue search (find the venue where the photograph was taken) and 2) group venue search (find relevant venues that have the same category as the photograph), by the cross-modal correlation between the input photograph and textual description of venues. In this model, data in different modalities are projected to a same space via deep networks. Pairwise correlation (between different modality data from the same venue) for exact venue search and category-based correlation (between different modality data from different venues with the same category) for group venue search are jointly optimized. Because a photograph cannot fully reflect rich text description of a venue, the number of photographs per venue in the training phase is increased to capture more aspects of a venue. We build a new venue-aware multimodal data set by integrating Wikipedia featured articles and Foursquare venue photographs. Experimental results on this data set confirm the feasibility of the proposed method. Moreover, the evaluation over another publicly available data set confirms that the proposed method outperforms state of the arts for cross-modal retrieval between image and text.

Context-Patch Face Hallucination Based on Thresholding Locality-Constrained Representation and Reproducing Learning

Face hallucination is a technique that reconstructs high-resolution (HR) faces from low-resolution (LR) faces, by using the prior knowledge learned from HR/LR face pairs. Most state-of-the-arts leverage position-patch prior knowledge of the human face to estimate the optimal representation coefficients for each image patch. However, they focus only the position information and usually ignore the context information of the image patch. In addition, when they are confronted with misalignment or the small sample size (SSS) problem, the hallucination performance is very poor. To this end, this paper incorporates the contextual information of the image patch and proposes a powerful and efficient context-patch-based face hallucination approach, namely, thresholding locality-constrained representation and reproducing learning (TLcR-RL). Under the context-patch-based framework, we advance a thresholding-based representation method to enhance the reconstruction accuracy and reduce the computational complexity. To further improve the performance of the proposed algorithm, we propose a promotion strategy called reproducing learning. By adding the estimated HR face to the training set, which can simulate the case that the HR version of the input LR face is present in the training set, it thus iteratively enhances the final hallucination result. Experiments demonstrate that the proposed TLcR-RL method achieves a substantial increase in the hallucinated results, both subjectively and objectively. In addition, the proposed framework is more robust to face misalignment and the SSS problem, and its hallucinated HR face is still very good when the LR test face is from the real world. The MATLAB source code is available at https://github.com/junjun-jiang/TLcR-RL.

Analysis of gene expression in mouse 2-cell embryos using fluorescein differential display: comparison of culture environments

The effect of the oviductal environment on gene expression in 2-cell mouse embryos was examined with mRNA differential display. Embryos used for experiments were cultured in modified Whitten medium with or without oviductal tissue until late 2-cell stage. The results of sequencing indicated that the genes for ATP synthase (ATPase 6), S:-adenosylmethionine decarboxylase (S:-AMDC) and nuclear autoantigenic sperm protein (NASP) were differentially expressed in embryos cultured in the oviductal environment (nonblocking culture condition). The ATPase 6 gene is encoded by mitochondrial DNA and is essential for the production of ATP. This indicates that the expression of ATP synthesis-related genes at the 2-cell stage may be required to maintain normal development in vitro. S:-Adenosylmethionine decarboxylase decarboxylates adenosylmethionine, which is a substrate of DNA methylation. The expression of S:-AMDC may be responsible for the low level of methylation of preimplantation development. As NASP is a histone-binding protein that is thought to be testis and sperm specific, its function in embryos remains unclear. On the other hand, the Tcl1 gene and a novel gene, the c-1 gene, were strongly expressed in embryos cultured without oviductal tissue (blocking culture condition). The expression patterns of these genes are quite similar. However, the detailed functions of these genes in embryos remain to be determined.

Structure, expression, and conserved physical linkage of mouse testicular cell adhesion molecule-1 (TCAM-1) gene

Isolation and characterization were performed for cDNA encoding mouse testicular cell adhesion molecule-1 (TCAM-1) using 2908 bases coding for a protein having 548 amino acids (60 kDa). Mouse TCAM-1 protein was found to consist of seven domains for signal sequence, five immunoglobulin (Ig) domains, and the transmembrane plus cytoplasmic domain. TCAM-1 gene and the region linking it to growth hormone (GH) gene located downstream from the TCAM-1 gene were then analyzed. The mouse TCAM-1 gene was 11.6 kb in length with 8 exons; the same as for the 12.0 kb rat gene. The distance from the TCAM-1 to GH gene was 12.5 kb in the mouse genome, and 7.6 kb in the rat. By Northern hybridization, 3.1-kb TCAM-1 mRNA was detected in 17-day testis and would appear present in pachytene spermatocytes and round spermatids.

Genomic analysis of the mouse protamine 1, protamine 2, and transition protein 2 gene cluster reveals hypermethylation in expressing cells

To understand the role of chromatin structure in the expression of the mouse protamine 1, protamine 2, and transition protein 2 genes during spermatogenesis, we have examined the genomic organization of this cluster of "haploid-specific" genes. As seen in the human genome, protamine 2, transition protein 2, and approximately 2.8 kb of a CpG island, hereafter called CpG island-dTP2, were clustered in a small region. Methylation analyses of this region have demonstrated that i) unlike most other tissue-specific genes, the protamine 1, protamine 2, and transition protein 2 genes were located in a large methylated domain in round spermatids, the cell type where they are transcribed, ii) the protamine 1 gene was only partially methylated in somatic cells and in testes from 7-day-old mice, and iii) the approximately 2 kb upstream and downstream of the CpG island-dTP2 were only partially methylated in somatic tissues. DNase I analysis revealed the presence of at least five strong DNase I hypersensitive sites over the CpG island-dTP2 in somatic tissues, but not in germ cells, and sequence analysis indicated that the CpG island-dTP2 is homologous to a CpG island located approximately 10.6 kb downstream of the human transition protein 2 gene. Although the nature of a CpG island-dTP2 and the function of a CpG island-dTP2-containing somatic tissue-specific DNase I hypersensitive sites in close proximity to the germ cell-specific gene cluster are unclear, the "open" chromatin structure of the CpG island-dTP2 may be responsible for the partial methylation pattern of the flanking sequences including the transition protein 2 gene in somatic tissues.

Significance of the chromatin structure in expression of the rat prolactin gene

To elucidate the mechanisms underlying cell type-specific expression of the growth hormone (GH) and prolactin (PRL) genes, we used rat pituitary-derived cell lines producing exclusively GH (GC cells) or PRL (235 cells), and examined the following: expression of transcription factors essential for GH and/or PRL gene expression; promoter/enhancer activity of the GH and PRL genes transiently introduced by transfection; and chromatin structures of the GH and PRL genes. Even in PRL-nonproducing GC cells, the PRL promoter/enhancer was more active than the GH promoter, and the transcription factors, Pit-1 and estrogen receptor (ER), essential for PRL gene expression were functional. The PRL promoter/enhancer of GC cells was normal. On DNase I sensitivity analysis of the chromatin structure, two hypersensitive sites were detected in PRL gene chromatin of PRL-producing 235 cells but none in that of GC cells. It thus follows that the major reason for absence of the expression of the endogenous PRL gene in GC cells is neither the lack of transcription factors necessary for PRL gene expression nor an anomaly of the PRL gene itself, but that the chromatin structure of the PRL gene differs in PRL-nonproducing and -producing cells. It was shown in this study that neither Pit-1 nor ER is required for conversion of the structure of PRL gene chromatin to a DNase I-hypersensitive state.

DNase I-hypersensitive sites in the chromatin of rat growth hormone gene locus and enhancer activity of regions with these sites

In this study, a determination was made of the chromatin structure of the rat growth hormone (GH) gene locus by DNase I sensitivity analysis using GC [GH+, prolactin (PRL)-], 235 (GH-, PRL+), GH3 (GH+, PRL+) and liver (GH-, PRL-) cells. From 7 kb upstream from the transcription start site to 19 kb downstream from the polyadenylation site, two major DNase I-hypersensitive sites (M-DHS; UIA, UIIA) and three M-DHS (DIA, DII, DIII) were found within 2 kb upstream and 7 kb downstream regions, respectively. Two minor DHS (m-DHS; UIB, UIIB) in the upstream region and one m-DHS (DIB) downstream were shown to be associated with M-DHS. Thus, a total of five M-DHS and three m-DHS were mapped on the rat GH gene locus. Among these, five (UIIB, UIA, UIB, DIB, DIA) including two (UIA, DIA) M-DHS were specific for GH-producing cells. UIIA and DIII were M-DHS only in PRL-producing 235 cells while the major hypersensitivity of DII was detected in GH-producing cells and liver cells. Assessment of the enhancing activity of the DHS regions indicated novel enhancers in one upstream and two downstream regions that function well with the GH promoter in GC cells. These enhancers, each appearing different, coincided with m-DHS but not M-DHS in GC cells, and were not activated by Pit-1. Based on these observations, the following functions of five M-DHS and three m-DHS regions were defined: enhancer; locus control region (LCR); switch region serving for conversion from GH/PRL-producing cells to PRL-producing cells; and a region having a structural function in chromatin.

The regulatory region and transcription factor required for the expression of rat and salmon pituitary hormone-encoding genes show cell-type and species specificity

The promoter regions of the genes encoding the rat and chum salmon growth hormones (GH) and rat prolactin (PRL) were combined with a reporter gene and introduced into GH- and/or PRL-producing cells from rat. The rat GH and PRL promoters (pGH and pPRL, respectively) were most active in cells producing GH and PRL, respectively. The activity of the salmon pGH was much less than that of the rat pGH in rat GH-producing cells. The regulatory region required for cell-type-specific gene expression of pituitary hormones thus contains information, not only for cell-type specificity, but possibly for species specificity as well. A reporter plasmid containing the GH or somatolactin (SL) promoter and an effector plasmid having a gene encoding transcription factor Pit-1 (rat or salmon) were cotransfected into HeLa (human) or EPC (carp) cells. Rat and salmon Pit-1 were more active in HeLa and EPC cells, respectively, indicating that Pit-1 appears to interact species specifically with the transcription machinery.

Effects of MK-733, an inhibitor of 3-hydroxy-3-methylglutaryl-coenzyme A reductase, on absorption and excretion of [3H]cholesterol in rabbits

Effects of MK-733 on the absorption and excretion of cholesterol in rabbits were examined using [3H]cholesterol. The animals were divided into six groups; three groups were fed a normal diet, and the other groups a cholesterol diet. MK-733 was administered orally as a single dose of 10 mg/kg on day 8, or multiple doses of 10 mg/kg once a day for 14 days. On the 8th day, [3H]cholesterol was given orally to each animal. In the groups fed a normal diet, single and consecutive administration of MK-733 did not affect the absorption and excretion of [3H]cholesterol. In the cholesterol-fed groups, however, single administration of MK-733 decreased the serum 3H radioactivity slightly, but did not affect the fecal excretion of [3H]cholesterol. However, the consecutive treatment with MK-733 clearly reduced the serum 3H radioactivity. The cumulative excretion of the fecal radioactivity of [3H]cholesterol in the MK-733 group (multiple) was higher than that in the control group. From these results, it is concluded that MK-733 inhibits the absorption of cholesterol from the gastrointestinal wall in cholesterol-fed rabbits.

Cytological diagnosis of leiomyogenic tumors of the stomach

The cytological findings in 7 cases of leiomyoma and 4 cases of leiomyosarcoma can be summarized as follows: For differential diagnosis of these lesions, in comparison with leiomyoma cells, leiomyosarcoma cells were found to have 1) increased minor axis diameters of the nuclei and nuclear anisokaryosis, 2) dark nuclear staining, 3) enlarged and darkly stained chromocenters, 4) dark staining and thickening of the nuclear rim, 5) enlargement of nuclear clear areas, 6) increased numbers of oval nuclei and greater pleomorphism, 7) an increase in size and number of the nucleoli, 8) a strong tendency for cell atypism, such as anisocytosis and pleomorphism. In the light of these findings, it is believed that differential diagnosis is indeed possible.

[Effect of ursodesoxycholic acid and cholic acid on intestinal absorption of cholesterol]

A study was carried out to investigate the effect of ursodesoxycholic acid and cholic acid on intestinal absorption of cholesterol-4-14C administered p.o. to lymph-fistula rats. The results indicated that within 24 hours, the labeled cholesterol detected in thoracic duct lymph of the control group was 21.0% of the administered cholesterol-4-14C, whereas the group treated with cholic acid (250 mg/kg p.o.) was found to have an increased value of 30.6% absorbed cholesterol. In comparison, the group treated with the same dose of ursodesoxycholic acid showed a decreased value of 12.1%, indicating an inhibitory effect on cholesterol absorption by the compound. This differential effect of the two compounds was also observed in a time-course study.