Development of a machine learning framework for radiation biomarker discovery and absorbed dose prediction.

Background: Molecular radiation biomarkers are an emerging tool in radiation research with applications for cancer radiotherapy, radiation risk assessment, and even human space travel. However, biomarker screening in genome-wide expression datasets using conventional tools is time-consuming and underlies analyst (human) bias. Machine Learning (ML) methods can improve the sensitivity and specificity of biomarker identification, increase analytical speed, and avoid multicollinearity and human bias. Aim: To develop a resource-efficient ML framework for radiation biomarker discovery using gene expression data from irradiated normal tissues. Further, to identify biomarker panels predicting radiation dose with tissue specificity. Methods: A strategic search in the Gene Expression Omnibus database identified a transcriptomic dataset (GSE44762) for normal tissues radiation responses (murine kidney cortex and medulla) suited for biomarker discovery using an ML approach. The dataset was pre-processed in R and separated into train and test data subsets. High computational cost of Genetic Algorithm/k-Nearest Neighbor (GA/KNN) mandated optimization and 13 ML models were tested using the caret package in R. Biomarker performance was evaluated and visualized via Principal Component Analysis (PCA) and dose regression. The novelty of ML-identified biomarker panels was evaluated by literature search. Results: Caret-based feature selection and ML methods vastly improved processing time over the GA approach. The KNN method yielded overall best performance values on train and test data and was implemented into the framework. The top-ranking genes were Cdkn1a, Gria3, Mdm2 and Plk2 in cortex, and Brf2, Ccng1, Cdkn1a, Ddit4l, and Gria3 in medulla. These candidates successfully categorized dose groups and tissues in PCA. Regression analysis showed that correlation between predicted and true dose was high with R2 of 0.97 and 0.99 for cortex and medulla, respectively. Conclusion: The caret framework is a powerful tool for radiation biomarker discovery optimizing performance with resource-efficiency for broad implementation in the field. The KNN-based approach identified Brf2, Ddit4l, and Gria3 mRNA as novel candidates that have been uncharacterized as radiation biomarkers to date. The biomarker panel showed good performance in dose and tissue separation and dose regression. Further training with larger cohorts is warranted to improve accuracy, especially for lower doses.

PΨFinder: a practical tool for the identification and visualization of novel pseudogenes in DNA sequencing data.

BACKGROUND: Processed pseudogenes (PΨgs) are disabled gene copies that are transcribed and may affect expression of paralogous genes. Moreover, their insertion in the genome can disrupt the structure or the regulatory region of a gene, affecting its expression level. These events have been identified as occurring mutations during cancer development, thus being able to identify PΨgs and their location will improve their impact on diagnostic testing, not only in cancer but also in inherited disorders. RESULTS: We have implemented PΨFinder (P-psy-finder), a tool that identifies PΨgs, annotates known ones and predicts their insertion site(s) in the genome. The tool screens alignment files and provides user-friendly summary reports and visualizations. To demonstrate its applicability, we scanned 218 DNA samples from patients screened for hereditary colorectal cancer. We detected 423 PΨgs distributed in 96% of the samples, comprising 7 different parent genes. Among these, we confirmed the well-known insertion site of the SMAD4-PΨg within the last intron of the SCAI gene in one sample. While for the ubiquitous CBX3-PΨg, present in 82.6% of the samples, we found it reversed inserted in the second intron of the C15ORF57 gene. CONCLUSIONS: PΨFinder is a tool that can automatically identify novel PΨgs from DNA sequencing data and determine their location in the genome with high sensitivity (95.92%). It generates high quality figures and tables that facilitate the interpretation of the results and can guide the experimental validation. PΨFinder is a complementary analysis to any mutational screening in the identification of disease-causing mutations within cancer and other diseases.

Comparison of online learning designs during the COVID-19 pandemic within bioinformatics courses in higher education.

MOTIVATION: Due to the worldwide COVID-19 pandemic, new strategies had to be adopted to move from classroom-based education to online education, in a very short time. The lack of time to set up these strategies, hindered a proper design of online instructions and delivery of knowledge. Bioinformatics-related training and other onsite practical education, tend to rely on extensive practice, where students and instructors have a face-to-face interaction to improve the learning outcome. For these courses to maintain their high quality when adapted as online courses, different designs need to be tested and the students' perceptions need to be heard. RESULTS: This study focuses on short bioinformatics-related courses for graduate students at the University of Gothenburg, Sweden, which were originally developed for onsite training. Once adapted as online courses, several modifications in their design were tested to obtain the best fitting learning strategy for the students. To improve the online learning experience, we propose a combination of: (i) short synchronized sessions, (ii) extended time for own and group practical work, (iii) recorded live lectures and (iv) increased opportunities for feedback in several formats. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Comparison of DNA Methylation Profiles of Hemostatic Genes between Liver Tissue and Peripheral Blood within Individuals.

DNA methylation has become increasingly recognized in the etiology of complex diseases, including thrombotic disorders. Blood is often collected in epidemiological studies for genotyping and has recently also been used to examine DNA methylation in epigenome-wide association studies. DNA methylation patterns are often tissue-specific, thus, peripheral blood may not accurately reflect the methylation pattern in the tissue of relevance. Here, we collected paired liver and blood samples concurrently from 27 individuals undergoing liver surgery. We performed targeted bisulfite sequencing for a set of 35 hemostatic genes primarily expressed in liver to analyze DNA methylation levels of >10,000 cytosine-phosphate-guanine (CpG) dinucleotides. We evaluated whether DNA methylation in blood could serve as a proxy for DNA methylation in liver at individual CpGs. Approximately 30% of CpGs were nonvariable and were predominantly hypo- (<25%) or hypermethylated (>70%) in both tissues. While blood can serve as a proxy for liver at these CpGs, the low variability renders these unlikely to explain phenotypic differences. We therefore focused on CpG sites with variable methylation levels in liver. The level of blood-liver tissue correlation varied widely across these variable CpGs; moderate correlations (0.5 ≤ r < 0.75) were detected for 6% and strong correlations (r ≥ 0.75) for a further 4%. Our findings indicate that it is essential to study the concordance of DNA methylation between blood and liver at individual CpGs. This paired blood-liver dataset is intended as a resource to aid interpretation of blood-based DNA methylation results.

RNA Sequencing of Hepatobiliary Cancer Cell Lines: Data and Applications to Mutational and Transcriptomic Profiling.

Cancer cell lines allow the identification of clinically relevant alterations and the prediction of drug response. However, sequencing data for hepatobiliary cancer cell lines in general, and particularly gallbladder cancer (GBC), are sparse. Here, we apply RNA sequencing to characterize 10 GBC, eight hepatocellular carcinoma, and five cholangiocarcinoma (CCA) cell lines. RNA extraction, quality control, library preparation, sequencing, and pre-processing of sequencing data were implemented using state-of-the-art techniques. Public data from the MSK-IMPACT database and a large cohort of Japanese biliary tract cancer patients were used to illustrate the usage of the released data. The total number of exonic mutations varied from 7207 for the cell line NOZ to 9760 for HuCCT1. Researchers planning experiments that require TP53 mutations could use the cell lines NOZ, OCUG-1, SNU308, or YoMi. Mz-Cha-1 showed mutations in ATM, SNU308 presented SMAD4 mutations, and the only investigated cell line that showed ARID1A mutations was GB-d1. SNU478 was the cell line with the global gene expression pattern most similar to GBC, intrahepatic CCA, and extrahepatic CCA. EGFR, KMT2D, and KMT2C generally presented a higher expression in the investigated cell lines than in Japanese primary GBC tumors. We provide the scientific community with detailed mutation and gene expression data, together with three showcase applications, with the aim of facilitating the design of future in vitro cell culture assays for research on hepatobiliary cancer.

Functional and genomic characterization of three novel cell lines derived from a metastatic gallbladder cancer tumor.

BACKGROUND: Gallbladder cancer (GBC) is the most common tumor of the biliary tract. The incidence of GBC shows a large geographic variability, being particularly frequent in Native American populations. In Chile, GBC represents the second cause of cancer-related death among women. We describe here the establishment of three novel cell lines derived from the ascitic fluid of a Chilean GBC patient, who presented 46% European, 36% Mapuche, 12% Aymara and 6% African ancestry. RESULTS: After immunocytochemical staining of the primary cell culture, we isolated and comprehensively characterized three independent clones (PUC-GBC1, PUC-GBC2 and PUC-GBC3) by short tandem repeat DNA profiling and RNA sequencing as well as karyotype, doubling time, chemosensitivity, in vitro migration capability and in vivo tumorigenicity assay. Primary culture cells showed high expression of CK7, CK19, CA 19-9, MUC1 and MUC16, and negative expression of mesothelial markers. The three isolated clones displayed an epithelial phenotype and an abnormal structure and number of chromosomes. RNA sequencing confirmed the increased expression of cytokeratin and mucin genes, and also of TP53 and ERBB2 with some differences among the three cells lines, and revealed a novel exonic mutation in NF1. The PUC-GBC3 clone was the most aggressive according to histopathological features and the tumorigenic capacity in NSG mice. CONCLUSIONS: The first cell lines established from a Chilean GBC patient represent a new model for studying GBC in patients of Native American descent.

A Comprehensive Sequencing-Based Analysis of Allelic Methylation Patterns in Hemostatic Genes in Human Liver.

Characterizing the relationship between genetic, epigenetic (e.g., deoxyribonucleic acid [DNA] methylation), and transcript variation could provide insights into mechanisms regulating hemostasis and potentially identify new drug targets. Several hemostatic factors are synthesized in the liver, yet high-resolution DNA methylation data from human liver tissue is currently lacking for these genes. Single-nucleotide polymorphisms (SNPs) can influence DNA methylation in cis which can affect gene expression. This can be analyzed through allele-specific methylation (ASM) experiments. We performed targeted genomic DNA- and bisulfite-sequencing of 35 hemostatic genes in human liver samples for SNP and DNA methylation analysis, respectively, and integrated the data for ASM determination. ASM-associated SNPs (ASM-SNPs) were tested for association to gene expression in liver using in-house generated ribonucleic acid-sequencing data. We then assessed whether ASM-SNPs associated with gene expression, plasma proteins, or other traits relevant for hemostasis using publicly available data. We identified 112 candidate ASM-SNPs. Of these, 68% were associated with expression of their respective genes in human liver or in other human tissues and 54% were associated with the respective plasma protein levels, activity, or other relevant hemostatic genome-wide association study traits such as venous thromboembolism, coronary artery disease, stroke, and warfarin dose maintenance. Our study provides the first detailed map of the DNA methylation landscape and ASM analysis of hemostatic genes in human liver tissue, and suggests that methylation regulated by genetic variants in cis may provide a mechanistic link between noncoding SNPs and variation observed in circulating hemostatic proteins, prothrombotic diseases, and drug response.

Functional and genomic characterization of three novel cell lines derived from a metastatic gallbladder cancer tumor

BACKGROUND: Gallbladder cancer (GBC) is the most common tumor of the biliary tract. The incidence of GBC shows a large geographic variability, being particularly frequent in Native American populations. In Chile, GBC represents the second cause of cancer-related death among women. We describe here the establishment of three novel cell lines derived from the ascitic fluid of a Chilean GBC patient, who presented 46% European, 36% Mapuche, 12% Aymara and 6% African ancestry. RESULTS: After immunocytochemical staining of the primary cell culture, we isolated and comprehensively characterized three independent clones (PUC-GBC1, PUC-GBC2 and PUC-GBC3) by short tandem repeat DNA profiling and RNA sequencing as well as karyotype, doubling time, chemosensitivity, in vitro migration capability and in vivo tumorigenicity assay. Primary culture cells showed high expression of CK7, CK19, CA 19-9, MUC1 and MUC16, and negative expression of mesothelial markers. The three isolated clones displayed an epithelial phenotype and an abnormal structure and number of chromosomes. RNA sequencing confirmed the increased expression of cytokeratin and mucin genes, and also of TP53 and ERBB2 with some differences among the three cells lines, and revealed a novel exonic mutation in NF1. The PUC-GBC3 clone was the most aggressive according to histopathological features and the tumorigenic capacity in NSG mice. CONCLUSIONS: The first cell lines established from a Chilean GBC patient represent a new model for studying GBC in patients of Native American descent.

Hemostatic Genes Exhibit a High Degree of Allele-Specific Regulation in Liver.

OBJECTIVE: Elucidating the genetic basis underlying hepatic hemostatic gene expression variability may contribute to unraveling genetic factors contributing to thrombotic or bleeding disorders. We aimed to identify novel cis-regulatory variants involved in regulating hemostatic genes by analyzing allele-specific expression (ASE) in human liver samples. STUDY DESIGN: Biopsies of human liver tissue and blood were collected from adults undergoing liver surgery at the Sahlgrenska University Hospital (n = 20). Genomic deoxyribonucleic acid (gDNA) and total ribonucleic acid (RNA) were isolated. A targeted approach was used to enrich and sequence 35 hemostatic genes for single nucleotide polymorphism (SNP) analysis (gDNAseq) and construct individualized genomes for transcript alignment. The allelic ratio of transcripts from targeted RNAseq was determined via ASE analysis. Public expression quantitative trait loci (eQTL) and genome-wide association study (GWAS) data were used to assess novelty and importance of the ASE SNPs (and proxies, r 2 ≥ 0.8) for relevant traits/diseases. RESULTS: Sixty percent of the genes studied showed allelic imbalance across 53 SNPs. Of these, 7 SNPs were previously validated in liver eQTL studies. For 32 with eQTLs in other cell/tissue types, this is the first time genotype-specific expression is demonstrated in liver, and for 14 ASE SNPs, this is the first ever reported genotype-expression association. A total of 29 ASE SNPs were previously associated with the respective plasma protein levels and 17 ASE SNPs to other relevant GWAS traits including venous thromboembolism, coronary artery disease, and stroke. CONCLUSION: Our study provides a comprehensive ASE analysis of hemostatic genes and insights into the regulation of hemostatic genes in human liver.

Hypoxia-regulated gene expression explains differences between melanoma cell line-derived xenografts and patient-derived xenografts.

Cell line-derived xenografts (CDXs) are an integral part of drug efficacy testing during development of new pharmaceuticals against cancer but their accuracy in predicting clinical responses in patients have been debated. Patient-derived xenografts (PDXs) are thought to be more useful for predictive biomarker identification for targeted therapies, including in metastatic melanoma, due to their similarities to human disease. Here, tumor biopsies from fifteen patients and ten widely-used melanoma cell lines were transplanted into immunocompromised mice to generate PDXs and CDXs, respectively. Gene expression profiles generated from the tumors of these PDXs and CDXs clustered into distinct groups, despite similar mutational signatures. Hypoxia-induced gene signatures and overexpression of the hypoxia-regulated miRNA hsa-miR-210 characterized CDXs. Inhibition of hsa-miR-210 with decoys had little phenotypic effect in vitro but reduced sensitivity to MEK1/2 inhibition in vivo, suggesting down-regulation of this miRNA could result in development of resistance to MEK inhibitors.

Regulation of DNA replication at the end of the mitochondrial D-loop involves the helicase TWINKLE and a conserved sequence element.

The majority of mitochondrial DNA replication events are terminated prematurely. The nascent DNA remains stably associated with the template, forming a triple-stranded displacement loop (D-loop) structure. However, the function of the D-loop region of the mitochondrial genome remains poorly understood. Using a comparative genomics approach we here identify two closely related 15 nt sequence motifs of the D-loop, strongly conserved among vertebrates. One motif is at the D-loop 5'-end and is part of the conserved sequence block 1 (CSB1). The other motif, here denoted coreTAS, is at the D-loop 3'-end. Both these sequences may prevent transcription across the D-loop region, since light and heavy strand transcription is terminated at CSB1 and coreTAS, respectively. Interestingly, the replication of the nascent D-loop strand, occurring in a direction opposite to that of heavy strand transcription, is also terminated at coreTAS, suggesting that coreTAS is involved in termination of both transcription and replication. Finally, we demonstrate that the loading of the helicase TWINKLE at coreTAS is reversible, implying that this site is a crucial component of a switch between D-loop formation and full-length mitochondrial DNA replication.

Polyglucosan body myopathy caused by defective ubiquitin ligase RBCK1.

Glycogen storage diseases are important causes of myopathy and cardiomyopathy. We describe 10 patients from 8 families with childhood or juvenile onset of myopathy, 8 of whom also had rapidly progressive cardiomyopathy, requiring heart transplant in 4. The patients were homozygous or compound heterozygous for missense or truncating mutations in RBCK1, which encodes for a ubiquitin ligase, and had extensive polyglucosan accumulation in skeletal muscle and in the heart in cases of cardiomyopathy. We conclude that RBCK1 deficiency is a frequent cause of polyglucosan storage myopathy associated with progressive muscle weakness and cardiomyopathy.

Histone modifications influence mediator interactions with chromatin.

The Mediator complex transmits activation signals from DNA bound transcription factors to the core transcription machinery. Genome wide localization studies have demonstrated that Mediator occupancy not only correlates with high levels of transcription, but that the complex also is present at transcriptionally silenced locations. We provide evidence that Mediator localization is guided by an interaction with histone tails, and that this interaction is regulated by their post-translational modifications. A quantitative, high-density genetic interaction map revealed links between Mediator components and factors affecting chromatin structure, especially histone deacetylases. Peptide binding assays demonstrated that pure wild-type Mediator forms stable complexes with the tails of Histone H3 and H4. These binding assays also showed Mediator-histone H4 peptide interactions are specifically inhibited by acetylation of the histone H4 lysine 16, a residue critical in transcriptional silencing. Finally, these findings were validated by tiling array analysis that revealed a broad correlation between Mediator and nucleosome occupancy in vivo, but a negative correlation between Mediator and nucleosomes acetylated at histone H4 lysine 16. Our studies show that chromatin structure and the acetylation state of histones are intimately connected to Mediator localization.

Analysis of gene order conservation in eukaryotes identifies transcriptionally and functionally linked genes.

The order of genes in eukaryotes is not entirely random. Studies of gene order conservation are important to understand genome evolution and to reveal mechanisms why certain neighboring genes are more difficult to separate during evolution. Here, genome-wide gene order information was compiled for 64 species, representing a wide variety of eukaryotic phyla. This information is presented in a browser where gene order may be displayed and compared between species. Factors related to non-random gene order in eukaryotes were examined by considering pairs of neighboring genes. The evolutionary conservation of gene pairs was studied with respect to relative transcriptional direction, intergenic distance and functional relationship as inferred by gene ontology. The results show that among gene pairs that are conserved the divergently and co-directionally transcribed genes are much more common than those that are convergently transcribed. Furthermore, highly conserved pairs, in particular those of fungi, are characterized by a short intergenic distance. Finally, gene pairs of metazoa and fungi that are evolutionary conserved and that are divergently transcribed are much more likely to be related by function as compared to poorly conserved gene pairs. One example is the ribosomal protein gene pair L13/S16, which is unusual as it occurs both in fungi and alveolates. A specific functional relationship between these two proteins is also suggested by the fact that they are part of the same operon in both eubacteria and archaea. In conclusion, factors associated with non-random gene order in eukaryotes include relative gene orientation, intergenic distance and functional relationships. It seems likely that certain pairs of genes are conserved because the genes involved have a transcriptional and/or functional relationship. The results also indicate that studies of gene order conservation aid in identifying genes that are related in terms of transcriptional control.

Conserved and variable domains of RNase MRP RNA.

Ribonuclease MRP is a eukaryotic ribonucleoprotein complex consisting of one RNA molecule and 7-10 protein subunits. One important function of MRP is to catalyze an endonucleolytic cleavage during processing of rRNA precursors. RNase MRP is evolutionary related to RNase P which is critical for tRNA processing. A large number of MRP RNA sequences that now are available have been used to identify conserved primary and secondary structure features of the molecule. MRP RNA has structural features in common with P RNA such as a conserved catalytic core, but it also has unique features and is characterized by a domain highly variable between species. Information regarding primary and secondary structure features is of interest not only in basic studies of the function of MRP RNA, but also because mutations in the RNA give rise to human genetic diseases such as cartilage-hair hypoplasia.

Computational screen for spliceosomal RNA genes aids in defining the phylogenetic distribution of major and minor spliceosomal components.

The RNA molecules of the spliceosome are critical for specificity and catalysis during splicing of eukaryotic pre-mRNA. In order to examine the evolution and phylogenetic distribution of these RNAs, we analyzed 149 eukaryotic genomes representing a broad range of phylogenetic groups. RNAs were predicted using high-sensitivity local alignment methods and profile HMMs in combination with covariance models. The results provide the most comprehensive view so far of the phylogenetic distribution of spliceosomal RNAs. RNAs were predicted in many phylogenetic groups where these RNA were not previously reported. Examples are RNAs of the major (U2-type) spliceosome in all fungal lineages, in lower metazoa and many protozoa. We also identified the minor (U12-type) spliceosomal U11 and U6atac RNAs in Acanthamoeba castellanii, where U12 spliceosomal RNA as well as minor introns were reported recently. In addition, minor-spliceosome-specific RNAs were identified in a number of phylogenetic groups where previously such RNAs were not observed, including the nematode Trichinella spiralis, the slime mold Physarum polycephalum and the fungal lineages Zygomycota and Chytridiomycota. The detailed map of the distribution of the U12-type RNA genes supports an early origin of the minor spliceosome and points to a number of occasions during evolution where it was lost.

Early evolution of histone mRNA 3' end processing.

The replication-dependent histone mRNAs in metazoa are not polyadenylated, in contrast to the bulk of mRNA. Instead, they contain an RNA stem-loop (SL) structure close to the 3' end of the mature RNA, and this 3' end is generated by cleavage using a machinery involving the U7 snRNP and protein factors such as the stem-loop binding protein (SLBP). This machinery of 3' end processing is related to that of polyadenylation as protein components are shared between the systems. It is commonly believed that histone 3' end processing is restricted to metazoa and green algae. In contrast, polyadenylation is ubiquitous in Eukarya. However, using computational approaches, we have now identified components of histone 3' end processing in a number of protozoa. Thus, the histone mRNA stem-loop structure as well as the SLBP protein are present in many different protozoa, including Dictyostelium, alveolates, Trypanosoma, and Trichomonas. These results show that the histone 3' end processing machinery is more ancient than previously anticipated and can be traced to the root of the eukaryotic phylogenetic tree. We also identified histone mRNAs from both metazoa and protozoa that are polyadenylated but also contain the signals characteristic of histone 3' end processing. These results provide further evidence that some histone genes are regulated at the level of 3' end processing to produce either polyadenylated RNAs or RNAs with the 3' end characteristic of replication-dependent histone mRNAs.