IGB VCF parsing flow

Key classes used:
Classes used for parsing and visualisation are all custom, there are no inbuilt classes used

1. VCF

• Responsible for reading, parsing, storing variants from VCF file
• Uses manual string splitting and regex based parsing to extract metadata, variant information

2. SeqSymmetry

• Stores standard variant metadata

3. BAMSym

• Represents insertions, deletions, and other structural changes.

4. GraphIntervalSym

• Stores numeric data from the VCF file.
• Used for graphing info/format fields in IGB.

No in-built libraries used for parsing, it is handled in manual way.

VCF (parsing and processing)

1. Loading the file:

• The VCF file is opened as an InputStream for processing.

2. Reading and extracting metadata:

• inputstream is wrapped in bufferreader for better text reading, to process line by line

3. Parsing data metadata and header

• so if starts with "##", it extracts metadata INFO, FILTER, FORMAT and stores metadata in maps (infoMap, filterMap, formatMap).
• The header line (#CHROM POS ... FORMAT SAMPLE1 SAMPLE2 ...) is manually extracted and sample names are stored in array
• uses lot of regex patterns to extract metadata

4. Parsing variant information

• Convert POS from 1-based (vcf) to 0-based IGB
• Extracts information like info fields, sample-specific genotype
• Crate SeqSymmetry object for each variant.(location, REF, ALT, quality, genotype, etc.).
• Extracts genotypes manually from FORMAT fields
• Convert to BAMSym obj format for Structural Variants(insertion/deletion) which uses CIGAR format (e.g., M for match, I for insertion, D for deletion) and ensures correct alignment visualization.

5. Data visualisation in IGB

• Variants are displayed as annotations on genomic tracks.
• Numeric INFO fields (like depth, allele frequency) are plotted as graphs.

Issues with current VCF parsing logic in IGB:

1. Parsing approach

• IGB manually goes through file line by line and uses lot of regex expressions for splitting and extracting metadata
• multiple loop iterations in extracting genotype data, fields

2. Data handling

• Store metadata and genotype information in seperate maps and lists instead of single object like VariantContext to store everything
• needs too many iterations to access different parts of data

3. Error handling

• weak error handling, it may fail when there is missing or misformatted data
• lot of manual checks

Loading population-level VCF:

• IGB loads entire file, complex logic for per-sample tracks and allele frequency graphs and do not handle region-based loading unlike IGV

Individual-specific information

• IGB loads entire file and have simple logic to handle single sample data