UCSC Genome Browser Basics: Tutorial 1 for Genetics Research

ClinVar is a public database where clinical laboratories and research groups submit genetic variants with their classifications and supporting evidence, enabling genetic counselors to review variant significance through multiple submitters' assertions, star ratings indicating classification confidence, and supporting data such as family segregation and phenotypic information; effective variant searches require proper nomenclature and may need wildcard queries, while gene-level searches reveal variant distribution patterns and classification conflicts that inform clinical interpretation according to ACMG guidelines.

Variant annotation adds functional information to variants by querying databases. Key databases include: (1) dbSNP - provides variant IDs (RS IDs), allele frequencies across populations, and basic variant information; (2) ClinVar - provides clinical significance information (pathogenic, benign, uncertain); (3) 1000 Genomes - provides population allele frequencies across different ancestral groups. Annotation tools like ANNOVAR extract information from these databases and add it to VCF files. This information helps prioritize variants for further study, identify disease-causing mutations, and understand population genetics.

Annotating GWAS top candidates with ClinVar data provides clinical context for identified variants. Download the latest ClinVar variant summary file and load it into R as a data frame. Filter ClinVar data to include only SNPs and ensure chromosome columns match data types. Perform inner joins on both chromosome and position to merge GWAS top candidates with ClinVar annotations. This integration enables assessment of whether statistically significant variants have known clinical implications.

Large-scale genetic databases like gnomAD (126,000 exomes, 15,000 whole genomes) provide allele frequencies showing how common variants are across populations. These frequencies are critical for clinical labs assessing whether variants cause disease. CLINVAR, the largest repository of genetic variants with clinical interpretations, received over 95,000 variants from the Understand Your Genome program, with 99% from healthy participants. This data helps balance information from sick patients with healthy population data, improving variant interpretation accuracy and enabling better diagnosis for all patients.

Variant normalization ensures consistent representation by left-aligning variants and using minimal nucleotide changes. Multi-allelic records require decomposition before annotation. Annotation software like ANNOVAR links variants to population databases (ExAC, gnomAD), conservation scores, and clinical databases (ClinVar, COSMIC). Population frequency analysis filters common benign variants, with ExAC/gnomAD providing allele frequencies from 60,000 to 123,000 individuals. This enables prioritization of rare variants more likely to be clinically significant. Annotation databases consume significant storage (80-90GB) and computational resources, requiring careful management in pipeline design.

Comparative genomics uses evolutionary conservation to predict variant effects. By comparing gene sequences across many species and creating multiple sequence alignments, we can identify positions that are conserved (unchanged) across evolution versus positions that vary freely. Positions that are absolutely conserved across millions of years of evolution likely perform critical functions, so variants at these positions are more likely to be pathogenic. Positions that vary freely across species are less likely to be functionally important. This information is used in variant effect predictors like SIFT (Sorting Intolerant From Tolerant), which analyze conservation patterns to predict whether a variant affects protein function.

Genome browsers organize data using customizable tracks that can be individually enabled or disabled. Available tracks include allele frequency data, gene expression patterns across tissues (colored differently for each tissue), and conservation scores between species (human, mouse, rat, dog, elephant, zebrafish, frog). Conservation tracks reveal evolutionary conservation of genomic regions. Researchers can hide irrelevant tracks to focus on specific information types. Human genomes have the most available data, while other species have progressively less annotation. Cross-species comparison helps identify conserved functional elements and detect species-specific innovations.

Comparative genomics examines evolutionary relationships, identifies genomic differences, derives gene function through homology, and locates conserved regions. Ensembl performs taxonomic comparisons (vertebrates, plants, metazoa) and pan-taxonomic analyses across representative species. Gene trees are constructed by classifying representative proteins into families, aligning them, and reconciling against species trees. Two homologous relationships exist: orthologs (speciation events, retain function) and paralogs (duplication events, may evolve new functions). Visualization includes colored clades, speciation (blue) and duplication (red) node coding, expandable subtrees, and protein alignment views. Ortholog databases display species sets, ortholog types, alignment percentages, conservation scores, and confidence levels. Whole genome alignments identify conserved functional regions (coding and non-coding) and support gene predictions. Ensembl provides pairwise alignments (two-species comparisons) and multiple alignments (predefined species sets like 91 eutherian mammals). Synteny represents conserved gene order between species. Users configure tracks via the cog icon, enabling conservation score visualization, multiple alignment tracks, BLAST alignment tracks for specific species comparisons, and side-by-side species region comparisons.

This extensive section covers advanced comparative genomics capabilities in Ensembl. It begins with ortholog analysis, explaining how to view ortholog relationships across species, filter tables by species name, and view pairwise sequence alignments between orthologs in Cluster W format. The section then transitions to whole genome alignments, explaining their purpose in identifying conserved functional regions and the two types: pairwise alignments (using BLASTZ or LASTZnet tools) and multiple alignments across predefined species sets (EPO analysis for fish, eutherian mammals, primates; MERCON for amniotes; Cactus for mouse strains). Practical demonstrations show how to add comparative genomics tracks, interpret conservation scores (positive for conserved regions, negative for unconserved), and view pairwise genome alignments between specific species. The section concludes with advanced features including the Alignments menu for selecting species to view human region alignments and the Region Comparison feature for side-by-side views of chromosomal overview, one megabase overview, and aligned images across species.

Comparative genomics uses evolutionary conservation to identify functional elements by leveraging natural selection's elimination of deleterious mutations. The core principle is that preserved positions across species alignments indicate functional importance. Key concepts include: (1) Branch length trade-off—more distant species provide more mutation events but less conservation; many closely related species maximize signal-to-noise ratio; (2) Substitution matrices like BLOSUM score exchanges based on observed frequencies using log likelihood ratios comparing related vs unrelated hypotheses; (3) Hidden Markov Models model conserved vs non-conserved states probabilistically; (4) Position weight matrices estimate conservation patterns directly from alignments, revealing degenerate positions where any nucleotide preserves function. These methods progressively move from simple nucleotide counting to sophisticated probabilistic models capturing selection pressures.

Identifying causal regulatory variants requires integrating multiple functional genomics evidence layers: (1) Position Weight Matrices (PWM) predict transcription factor binding affinity based on sequence context; (2) Chromatin accessibility (DNase-seq) identifies open chromatin regions where TFs can bind; (3) Histone modifications (e.g., H3K4me3 for enhancers) mark regulatory regions; (4) Chromatin looping brings distant enhancers and target genes together in 3D space. Major resources include ENCODE, Roadmap Epigenomics, and GTEx projects providing genome-wide data across tissues and conditions.

Modern functional genomics integrates diverse molecular data: eQTLs (gene expression), sQTLs (splicing), methylation QTLs, histone modifications, and chromatin accessibility. CISO detects splicing QTLs using seesaw analysis and LiftCutter, revealing substantial eQTL-sQTL overlaps. Opera jointly analyzes GWAS with multiple XQTL datasets, achieving ~50% locus mapping. MAGIC extends this to over 400 datasets across seven omics levels, achieving 87% mapping success. These methods address the challenge that bulk eQTL data cannot distinguish cell types, requiring cell state deconvolution approaches.

The ENCODE consortium generates thousands of genomic experiments including DNase-seq, ChIP-seq, and histone mark assays. The core challenge is integrating these diverse datasets to annotate functional genomic regions. The approach involves creating DNase master peaks—unique, non-overlapping peaks spanning ~20% of the genome—by merging ENCODE and Roadmap data. These peaks are classified as proximal (within 2kb of TSS) or distal. Annotation combines intersection with transcription factor peaks and histone mark enrichment analysis (H3K4me3, H3K9ac, H3K27ac, H3K4me1), yielding candidate promoters and enhancers. Annotations are accessible via ENCODE portal, UCSC, and WashU browsers, with cell type/tissue provenance information. Future directions include open-source implementation, mouse annotations, improved TF data integration, and incorporation of 3D contact data.

Ensembl integrates regulatory data from sources like ENCODE and Roadmap Epigenomics to identify genomic regions with promoter or enhancer activity through computational analysis of histone modifications, transcription factor binding, and open chromatin signals; users can access this data by navigating to specific genomic regions or genes, configuring tracks to view multi-cell regulatory features, cell-line specific features, and raw ChIP-seq data, with features identified by their RSxxxxx IDs and displayed using color-coded tracks that show which cell lines exhibit regulatory activity at each genomic position.

This module demonstrates how to use the genome browser to analyze transcription factor binding to specific genomic regions. The workflow involves entering a gene of interest (MCCAN), simplifying the view by merging transcripts by symbol, and examining the ENCODE Transcription Factor Clustered track containing approximately 340 transcription factors profiled across multiple cell types. Users learn to activate tracks, enable transcription factor names, adjust visualization height for readability, and navigate between different genomic loci (such as TERT) while maintaining track persistence. The visualization shows transcription factors binding at transcription start sites, which is expected biological behavior.