3D convolutional neural networks uncover modality-specific brain-imaging predictors for Alzheimer's disease sub-scores.

Different aspects of cognitive functions are affected in patients with Alzheimer's disease. To date, little is known about the associations between features from brain-imaging and individual Alzheimer's disease (AD)-related cognitive functional changes. In addition, how these associations differ among different imaging modalities is unclear. Here, we trained and investigated 3D convolutional neural network (CNN) models that predicted sub-scores of the 13-item Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog13) based on MRI and FDG-PET brain-imaging data. Analysis of the trained network showed that each key ADAS-Cog13 sub-score was associated with a specific set of brain features within an imaging modality. Furthermore, different association patterns were observed in MRI and FDG-PET modalities. According to MRI, cognitive sub-scores were typically associated with structural changes of subcortical regions, including amygdala, hippocampus, and putamen. Comparatively, according to FDG-PET, cognitive functions were typically associated with metabolic changes of cortical regions, including the cingulated gyrus, occipital cortex, middle front gyrus, precuneus cortex, and the cerebellum. These findings brought insights into complex AD etiology and emphasized the importance of investigating different brain-imaging modalities.

Improving brain age estimates with deep learning leads to identification of novel genetic factors associated with brain aging.

To study genetic factors associated with brain aging, we first need to quantify brain aging. Statistical models have been created for estimating the apparent age of the brain, or predicted brain age (PBA), using imaging data. Recent studies have refined these models to obtain a more accurate PBA, but research has yet to demonstrate the scientific value of doing so. Here, we show that a more accurate PBA leads to better characterization of genetic factors associated with brain aging. We trained a convolutional neural network (CNN) model on 16,998 UK Biobank subjects to derive PBA, then conducted a genome-wide association study on the PBA, in which we identified single nucleotide polymorphisms from four independent loci significantly associated with brain aging, three of which were novel. By comparing association results based on the CNN-derived PBA to those based on a linear regression-derived PBA, we concluded that a more accurate PBA enables the discovery of novel genetic associations. Our results may be valuable for identifying other lifestyle factors associated with brain aging.

Identification of Dysregulated Genes for Late-Onset Alzheimer's Disease Using Gene Expression Data in Brain.

BACKGROUND: Alzheimer's Disease (AD) is a neurodegenerative complex brain disease that represents a public health concern. AD is considered the fifth leading cause of death in Americans who are older than 65 years which prioritizes the importance of understanding the etiology of AD in its early stages before the onset of symptoms. This study attempted to further understand Alzheimer's disease (AD) etiology by investigating the dysregulated genes using gene expression data from multiple brain regions. METHODS: A linear mixed-effects model for differential gene expression analysis was used in a sample of 15 AD and 30 control subjects, each with data from four different brain regions, in order to deal with the hierarchical multilevel data. Post-hoc Gene Ontology and pathway enrichment analyses provided insights on the biological implications in AD progression. Supervised machine learning algorithms were used to assess the discriminative power of the top 10 candidate genes in distinguishing between the two groups. RESULTS: Enrichment analyses revealed biological processes and pathways that are related to structural constituents and organization of the axons and synapses. These biological processes and pathways imply dysfunctional axon and synaptic transmission between neuronal cells in AD. Random Forest classification algorithm gave the best accuracy on the test data with F1-score of 0.88. CONCLUSION: The differentially expressed genes were associated with axon and synaptic transmissions which affect the neuronal connectivity in cognitive systems involved in AD pathophysiology. These genes may open ways to explore new effective treatments and early diagnosis before the onset of clinical symptoms.

Parity is associated with cognitive function and brain age in both females and males.

Previous studies of the association between parity and long-term cognitive changes have primarily focused on women and have shown conflicting results. We investigated this association by analyzing data collected on 303,196 subjects from the UK Biobank. We found that in both females and males, having offspring was associated with a faster response time and fewer mistakes made in the visual memory task. Subjects with two or three children had the largest differences relative to those who were childless, with greater effects observed in men. We further analyzed the association between parity and relative brain age (n = 13,584), a brain image-based biomarker indicating how old one's brain structure appears relative to peers. We found that in both sexes, subjects with two or three offspring had significantly reduced brain age compared to those without offspring, corroborating our cognitive function results. Our findings suggest that lifestyle factors accompanying having offspring, rather than the physical process of pregnancy experienced only by females, contribute to these associations and underscore the importance of studying such factors, particularly in the context of sex.

Association of relative brain age with tobacco smoking, alcohol consumption, and genetic variants.

Brain age is a metric that quantifies the degree of aging of a brain based on whole-brain anatomical characteristics. While associations between individual human brain regions and environmental or genetic factors have been investigated, how brain age is associated with those factors remains unclear. We investigated these associations using UK Biobank data. We first trained a statistical model for obtaining relative brain age (RBA), a metric describing a subject's brain age relative to peers, based on whole-brain anatomical measurements, from training set subjects (n = 5,193). We then applied this model to evaluation set subjects (n = 12,115) and tested the association of RBA with tobacco smoking, alcohol consumption, and genetic variants. We found that daily or almost daily consumption of tobacco and alcohol were both significantly associated with increased RBA (P < 0.001). We also found SNPs significantly associated with RBA (p-value < 5E-8). The SNP most significantly associated with RBA is located in MAPT gene. Our results suggest that both environmental and genetic factors are associated with structural brain aging.

Interaction effect of alcohol consumption and Alzheimer disease polygenic risk score on the brain cortical thickness of cognitively normal subjects.

Alcohol consumption and genetic risk for Alzheimer disease (AD) are among many factors known to be associated with brain structure in cognitively healthy adults. It is unclear, however, whether the effect of alcohol consumption on brain structure varies depending on a person's level of genetic risk for AD. We hypothesized that there is an interaction effect of alcohol consumption and a 33-SNP AD polygenic risk score (PRS) on the cortical thickness of brain regions known to be affected early in the course of AD. Studying 6,213 cognitively healthy subjects from the UK Biobank, we found a significant interaction effect of the 33-SNP AD PRS and alcohol consumption on this AD Cortical Thickness Signature. Stratified, among those who consume 12-24 g/day of alcohol, the 33-SNP AD PRS had a significant, positive association with AD Cortical Thickness Signature, with high-risk subjects having the greatest AD Cortical Thickness Signature. There were no significant associations of the 33-SNP AD PRS with AD Cortical Thickness Signature among the nondrinker or <1, 1-6, 6-12, 24-48, or >48 g/day groups. It is unclear whether this interaction is due to a detrimental or beneficial effect of moderate alcohol consumption in those with the highest genetic risk for AD.

Age-Related Differences in Brain Morphology and the Modifiers in Middle-Aged and Older Adults.

Brain structural morphology differs with age. This study examined age-differences in surface-based morphometric measures of cortical thickness, volume, and surface area in a well-defined sample of 8137 generally healthy UK Biobank participants aged 45-79 years. We illustrate that the complexity of age-related brain morphological differences may be related to the laminar organization and regional evolutionary history of the cortex, and age of about 60 is a break point for increasing negative associations between age and brain morphology in Alzheimer's disease (AD)-prone areas. We also report novel relationships of age-related cortical differences with individual factors of sex, cognitive functions of fluid intelligence, reaction time and prospective memory, cigarette smoking, alcohol consumption, sleep disruption, genetic markers of apolipoprotein E, brain-derived neurotrophic factor, catechol-O-methyltransferase, and several genome-wide association study loci for AD and further reveal joint effects of cognitive functions, lifestyle behaviors, and education on age-related cortical differences. These findings provide one of the most extensive characterizations of age associations with major brain morphological measures and improve our understanding of normal structural brain aging and its potential modifiers.

Classifying Alzheimer's disease with brain imaging and genetic data using a neural network framework.

A long-standing question is how to best use brain morphometric and genetic data to distinguish Alzheimer's disease (AD) patients from cognitively normal (CN) subjects and to predict those who will progress from mild cognitive impairment (MCI) to AD. Here, we use a neural network (NN) framework on both magnetic resonance imaging-derived quantitative structural brain measures and genetic data to address this question. We tested the effectiveness of NN models in classifying and predicting AD. We further performed a novel analysis of the NN model to gain insight into the most predictive imaging and genetics features and to identify possible interactions between features that affect AD risk. Data were obtained from the AD Neuroimaging Initiative cohort and included baseline structural MRI data and single nucleotide polymorphism (SNP) data for 138 AD patients, 225 CN subjects, and 358 MCI patients. We found that NN models with both brain and SNP features as predictors perform significantly better than models with either alone in classifying AD and CN subjects, with an area under the receiver operating characteristic curve (AUC) of 0.992, and in predicting the progression from MCI to AD (AUC=0.835). The most important predictors in the NN model were the left middle temporal gyrus volume, the left hippocampus volume, the right entorhinal cortex volume, and the APOE (a gene that encodes apolipoprotein E) ɛ4 risk allele. Furthermore, we identified interactions between the right parahippocampal gyrus and the right lateral occipital gyrus, the right banks of the superior temporal sulcus and the left posterior cingulate, and SNP rs10838725 and the left lateral occipital gyrus. Our work shows the ability of NN models to not only classify and predict AD occurrence but also to identify important AD risk factors and interactions among them.

A Frameshift in CSF2RB Predominant Among Ashkenazi Jews Increases Risk for Crohn's Disease and Reduces Monocyte Signaling via GM-CSF.

BACKGROUND & AIMS: Crohn's disease (CD) has the highest prevalence in Ashkenazi Jewish populations. We sought to identify rare, CD-associated frameshift variants of high functional and statistical effects. METHODS: We performed exome sequencing and array-based genotype analyses of 1477 Ashkenazi Jewish individuals with CD and 2614 Ashkenazi Jewish individuals without CD (controls). To validate our findings, we performed genotype analyses of an additional 1515 CD cases and 7052 controls for frameshift mutations in the colony-stimulating factor 2-receptor ß common subunit gene (CSF2RB). Intestinal tissues and blood samples were collected from patients with CD; lamina propria leukocytes were isolated and expression of CSF2RB and granulocyte-macrophage colony-stimulating factor-responsive cells were defined by adenomatous polyposis coli (APC) time-of-flight mass cytometry (CyTOF analysis). Variants of CSF2RB were transfected into HEK293 cells and the expression and functions of gene products were compared. RESULTS: In the discovery cohort, we associated CD with a frameshift mutation in CSF2RB (P = 8.52 × 10(-4)); the finding was validated in the replication cohort (combined P = 3.42 × 10(-6)). Incubation of intestinal lamina propria leukocytes with granulocyte-macrophage colony-stimulating factor resulted in high levels of phosphorylation of signal transducer and activator of transcription (STAT5) and lesser increases in phosphorylation of extracellular signal-regulated kinase and AK straining transforming (AKT). Cells co-transfected with full-length and mutant forms of CSF2RB had reduced pSTAT5 after stimulation with granulocyte-macrophage colony-stimulating factor, compared with cells transfected with control CSF2RB, indicating a dominant-negative effect of the mutant gene. Monocytes from patients with CD who were heterozygous for the frameshift mutation (6% of CD cases analyzed) had reduced responses to granulocyte-macrophage colony-stimulating factor and markedly decreased activity of aldehyde dehydrogenase; activity of this enzyme has been associated with immune tolerance. CONCLUSIONS: In a genetic analysis of Ashkenazi Jewish individuals, we associated CD with a frameshift mutation in CSF2RB. Intestinal monocytes from carriers of this mutation had reduced responses to granulocyte-macrophage colony-stimulating factor, providing an additional mechanism for alterations to the innate immune response in individuals with CD.

Improved integrative framework combining association data with gene expression features to prioritize Crohn's disease genes.

Genome-wide association studies in Crohn's disease (CD) have identified 140 genome-wide significant loci. However, identification of genes driving association signals remains challenging. Furthermore, genome-wide significant thresholds limit false positives at the expense of decreased sensitivity. In this study, we explored gene features contributing to CD pathogenicity, including gene-based association data from CD and autoimmune (AI) diseases, as well as gene expression features (eQTLs, epigenetic markers of expression and intestinal gene expression data). We developed an integrative model based on a CD reference gene set. This integrative approach outperformed gene-based association signals alone in identifying CD-related genes based on statistical validation, gene ontology enrichment, differential expression between M1 and M2 macrophages and a validation using genes causing monogenic forms of inflammatory bowel disease as a reference. Besides gene-level CD association P-values, association with AI diseases was the strongest predictor, highlighting generalized mechanisms of inflammation, and the interferon-γ pathway particularly. Within the 140 high-confidence CD regions, 598 of 1328 genes had low prioritization scores, highlighting genes unlikely to contribute to CD pathogenesis. For select regions, comparably high integrative model scores were observed for multiple genes. This is particularly evident for regions having extensive linkage disequilibrium such as the IBD5 locus. Our analyses provide a standardized reference for prioritizing potential CD-related genes, in regions with both highly significant and nominally significant gene-level association P-values. Our integrative model may be particularly valuable in prioritizing rare, potentially private, missense variants for which genome-wide evidence for association may be unattainable.

Pattern recognition receptor signaling in human dendritic cells is enhanced by ICOS ligand and modulated by the Crohn's disease ICOSLG risk allele.

Inflammatory bowel disease (IBD) is characterized by dysregulated intestinal immune homeostasis and cytokine secretion. Multiple loci are associated with IBD, but a functional explanation is missing for most. Here we found that pattern-recognition receptor (PRR)-induced cytokine secretion was diminished in human monocyte-derived dendritic cells (MDDC) from rs7282490 ICOSLG GG risk carriers. Homotypic interactions between the costimulatory molecule ICOS and the ICOS ligand on MDDCs amplified nucleotide-binding oligomerization domain 2 (NOD2)-initiated cytokine secretion. This amplification required arginine residues in the ICOSL cytoplasmic tail that recruited the adaptor protein RACK1 and the kinases PKC and JNK leading to PKC, MAPK, and NF-κB activation. MDDC from rs7282490 GG risk-carriers had reduced ICOSL expression and PRR-initiated signaling and this loss-of-function ICOSLG risk allele associated with an ileal Crohn's disease phenotype, similar to polymorphisms in NOD2. Taken together, ICOSL amplifies PRR-initiated outcomes, which might contribute to immune homeostasis.

Granulocyte-macrophage colony-stimulating factor autoantibodies: a marker of aggressive Crohn's disease.

BACKGROUND: Neutralizing autoantibodies (Abs) against granulocyte-macrophage colony-stimulating factor (GM-CSF Ab) have been associated with stricturing ileal Crohn's disease (CD) in a largely pediatric patient cohort (total 394, adult CD 57). The aim of this study was to examine this association in 2 independent predominantly adult inflammatory bowel disease patient cohorts. METHODS: Serum samples from 742 subjects from the NIDDK IBD Genetics Consortium and 736 subjects from Australia were analyzed for GM-CSF Ab and genetic markers. We conducted multiple regression analysis with backward elimination to assess the contribution of GM-CSF Ab levels and established CD risk alleles and smoking on ileal disease location in the 477 combined CD subjects from both cohorts. We also determined associations of GM-CSF Ab levels with complications requiring surgical intervention in combined CD subjects in both cohorts. RESULTS: Serum samples from patients with CD expressed significantly higher concentrations of GM-CSF Ab when compared with ulcerative colitis or controls in each cohort. Nonsmokers with ileal CD expressed significantly higher GM-CSF Ab concentrations in the Australian cohort (P = 0.002). Elevated GM-CSF Ab, ileal disease location, and disease duration more than 3 years were independently associated with stricturing/penetrating behavior and intestinal resection for CD. CONCLUSIONS: The expression of high GM-CSF Ab is a risk marker for aggressive CD behavior and complications including surgery. Modifying factors include environmental exposure to smoking and genetic risk markers.

Conserved epitopes dominate cross-CD8+ T-cell responses against influenza A H1N1 virus among Asian populations.

Novel strains of influenza A viruses (IAVs) have emerged with high infectivity and/or pathogenicity in recent years, causing worldwide concern. T cells are correlated with protection in humans through cross-reactive immunity against heterosubtypes of IAV. However, the different hierarchical roles of IAV-derived epitopes with distinct levels of polymorphism in the cross-reactive T-cell responses against IAV remain elusive. In this study, immunodominant epitopes scattered throughout the entire proteome of 2009 pandemic influenza A H1N1 virus and seasonal IAVs were synthesized and divided into different pools depending on their conservation. The overall profile of the IAV-specific CD8(+) T-cell immunity was detected by utilizing these peptide pools and also individual peptides. A dominant role of the conserved peptide-specific T-cell immunity was illuminated within the anti-IAV responses, while the CD8(+) T-cell responses against the variable epitopes were lower than the conserved peptides. As previously demonstrated within a Caucasian population, we determined that GL9-specific T cells, which also utilize Vß 17 TCR (BV19), play a pivotal role in IAV-specific T-cell immunity within an HLA-A2(+) Asian population. Our study objectively reveals the different predominant roles of T-cell epitopes among IAV-specific cross-reactive cellular immunity. This may guide the development of vaccines with cross-T-cell immunogenicity against heterosubtypes of IAV.

Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease.

Crohn's disease and ulcerative colitis, the two common forms of inflammatory bowel disease (IBD), affect over 2.5 million people of European ancestry, with rising prevalence in other populations. Genome-wide association studies and subsequent meta-analyses of these two diseases as separate phenotypes have implicated previously unsuspected mechanisms, such as autophagy, in their pathogenesis and showed that some IBD loci are shared with other inflammatory diseases. Here we expand on the knowledge of relevant pathways by undertaking a meta-analysis of Crohn's disease and ulcerative colitis genome-wide association scans, followed by extensive validation of significant findings, with a combined total of more than 75,000 cases and controls. We identify 71 new associations, for a total of 163 IBD loci, that meet genome-wide significance thresholds. Most loci contribute to both phenotypes, and both directional (consistently favouring one allele over the course of human history) and balancing (favouring the retention of both alleles within populations) selection effects are evident. Many IBD loci are also implicated in other immune-mediated disorders, most notably with ankylosing spondylitis and psoriasis. We also observe considerable overlap between susceptibility loci for IBD and mycobacterial infection. Gene co-expression network analysis emphasizes this relationship, with pathways shared between host responses to mycobacteria and those predisposing to IBD.

PICS: probabilistic inference for ChIP-seq.

ChIP-seq combines chromatin immunoprecipitation with massively parallel short-read sequencing. While it can profile genome-wide in vivo transcription factor-DNA association with higher sensitivity, specificity, and spatial resolution than ChIP-chip, it poses new challenges for statistical analysis that derive from the complexity of the biological systems characterized and from variability and biases in its sequence data. We propose a method called PICS (Probabilistic Inference for ChIP-seq) for identifying regions bound by transcription factors from aligned reads. PICS identifies binding event locations by modeling local concentrations of directional reads, and uses DNA fragment length prior information to discriminate closely adjacent binding events via a Bayesian hierarchical t-mixture model. It uses precalculated, whole-genome read mappability profiles and a truncated t-distribution to adjust binding event models for reads that are missing due to local genome repetitiveness. It estimates uncertainties in model parameters that can be used to define confidence regions on binding event locations and to filter estimates. Finally, PICS calculates a per-event enrichment score relative to a control sample, and can use a control sample to estimate a false discovery rate. Using published GABP and FOXA1 data from human cell lines, we show that PICS' predicted binding sites were more consistent with computationally predicted binding motifs than the alternative methods MACS, QuEST, CisGenome, and USeq. We then use a simulation study to confirm that PICS compares favorably to these methods and is robust to model misspecification.

BNArray: an R package for constructing gene regulatory networks from microarray data by using Bayesian network.

UNLABELLED: BNArray is a systemized tool developed in R. It facilitates the construction of gene regulatory networks from DNA microarray data by using Bayesian network. Significant sub-modules of regulatory networks with high confidence are reconstructed by using our extended sub-network mining algorithm of directed graphs. BNArray can handle microarray datasets with missing data. To evaluate the statistical features of generated Bayesian networks, re-sampling procedures are utilized to yield collections of candidate 1st-order network sets for mining dense coherent sub-networks. AVAILABILITY: The R package and the supplementary documentation are available at http://www.cls.zju.edu.cn/binfo/BNArray/.