When you get your whole genome sequenced, the provider doesn't just hand you a PDF of health reports. The real deliverable — the one that makes WGS worth the investment — is a set of raw data files that contain your complete genetic information in machine-readable format. These files are the permanent, re-analyzable dataset that gains value every year as genomic science advances.
The three files are FASTQ, BAM, and VCF. They represent three stages of the same data, from rawest to most refined. Think of them as the raw footage, the edited timeline, and the highlight reel of your genome. Here's exactly what each one is, what it looks like inside, and what it's good for.
FASTQ is the rawest form of your sequencing data. It contains every individual "read" — each short fragment of DNA that the sequencer detected — along with a quality score for each base. These reads are raw; they haven't been mapped to a reference genome yet, and they're not in chromosomal order. They're just millions of short DNA fragments, typically 100–150 base pairs long, exactly as they came off the Illumina NovaSeq sequencer.
Each read in a FASTQ file has four lines: a unique identifier, the DNA sequence itself (a string of A, T, C, and G letters), a separator, and an ASCII-encoded quality score for each base position. The quality score tells you how confident the sequencer was about each individual letter.
# What a single read looks like inside a FASTQ file: @SRR123456.1 HWI-D00523:240:HF3WGBCX2:1:1101:1203:2085 1:N:0:ATCACG NGATCTGGAGTTCAAATTCAGGCATGATTTCAGTTCTGGAAACTCATGT + #<<FFFFFFFIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII # Line 1: Read identifier (unique ID for this fragment) # Line 2: The actual DNA sequence (ATCG letters) # Line 3: Separator (always a +) # Line 4: Quality scores (ASCII characters, higher = better)
A 30× whole genome FASTQ dataset contains roughly 600–900 million reads. That's why these files are enormous.
BAM is what you get when you take those millions of FASTQ reads and align them to a reference human genome (typically GRCh38/hg38). Each read is now mapped to its exact position on a specific chromosome. The reads are sorted by chromosomal coordinate, indexed for fast random access, and stored in a compressed binary format.
A BAM file contains the same read data as a FASTQ file, plus alignment information: which chromosome each read maps to, what position, whether the read aligns perfectly or has mismatches, and whether the read maps to multiple locations. This is the format that variant-calling algorithms (like GATK HaplotypeCaller) use as input to identify where your DNA differs from the reference.
You can't open a BAM file in a text editor — it's binary. To view the contents, you need tools like SAMtools (command line) or IGV (Integrative Genomics Viewer, a graphical browser).
# Viewing a BAM file with SAMtools (shows aligned reads): $ samtools view your_genome.bam chr7:117,480,000-117,480,200 # Output: each line is one read aligned to chromosome 7 # Fields include: read name, flags, chromosome, position, # mapping quality, CIGAR string, sequence, and quality
VCF is the most compact and most commonly used of the three formats. It contains only the positions where your DNA differs from the reference genome — your variants. Each line in a VCF file describes one variant: which chromosome it's on, the exact position, what the reference base is, what your base is, and a quality/confidence score.
A typical human genome has roughly 4–5 million variants compared to the reference. That's a lot of differences, but it's a tiny fraction of the 6.4 billion total positions, which is why VCF files are orders of magnitude smaller than FASTQ or BAM. This is the file that consumer analysis platforms (SelfDecode, Promethease, FoundMyFitness) accept and work with.
# What a VCF file looks like inside: #CHROM POS ID REF ALT QUAL FILTER INFO chr1 10177 rs367896724 A AC 100 PASS ... chr1 10352 rs555500075 T TA 100 PASS ... chr7 117480025 rs113993960 ATCT A 99.9 PASS ... chr17 7674220 rs28934578 G A 85.3 PASS ... # Each line = one position where you differ from reference # CHROM: chromosome | POS: position | REF: reference base # ALT: your base | QUAL: confidence score
The Pipeline: How These Files Relate
These three formats aren't alternatives — they're sequential stages of the same analysis pipeline. The sequencer produces FASTQ. An alignment algorithm (like BWA-MEM2) maps those reads to the reference genome and produces BAM. A variant-calling algorithm (like GATK HaplotypeCaller) identifies positions where your reads differ from the reference and produces VCF.
Sequencer → FASTQ → Alignment (BWA-MEM2) → BAM → Variant Calling (GATK) → VCF
When your WGS provider delivers all three files, they've run this pipeline for you. The VCF is the end product you'll use most often. The BAM is useful for deeper analysis and clinical validation. The FASTQ is your insurance policy — the rawest possible data that lets you re-run the entire pipeline from scratch using any alignment algorithm, any reference genome, and any variant caller. If a significantly improved reference genome is released in 2028, you can re-align your FASTQ reads to it and potentially discover variants that were missed with the 2026 reference.
Which Files Do You Actually Need?
For consumer analysis (SelfDecode, Promethease, FoundMyFitness): VCF only. These platforms extract specific SNPs from your VCF and cross-reference them against health, ancestry, and pharmacogenomic databases. You don't need BAM or FASTQ for this.
For clinical or research use: BAM and VCF. A genetic counselor or clinical geneticist will want the BAM file to validate specific variant calls — they can visually inspect the reads supporting each variant in a genome browser like IGV. Some clinical WGS workflows also use BAM files to detect structural variants and copy-number changes.
For maximum future-proofing: All three. FASTQ files give you complete independence from any specific reference genome or analysis pipeline. As bioinformatics methods improve, you can re-process your raw reads from scratch. This is what biohackers, researchers, and data-first users want.
Storage and Backup Recommendations
Combined, a full set of FASTQ + BAM + VCF files from a 30× WGS run can exceed 150GB. That's not going to fit on most cloud storage free tiers or USB drives. Here's a practical approach:
VCF: Keep copies everywhere. It's small (under 1GB), and it's the file you'll use most. Cloud storage, local drive, email it to yourself — redundancy is cheap at this size.
BAM: One copy on a local external drive is sufficient for most people. Download it promptly after receiving results — don't assume indefinite access to your provider's cloud platform.
FASTQ: One copy on a large external drive or NAS. These files are your ultimate backup — the raw reads from which everything else can be regenerated. Worth preserving even if you never personally analyze them, because future bioinformatics tools may extract insights that current methods can't.
Get All Three: FASTQ + BAM + VCF
Dante Labs is one of the only consumer WGS providers that includes complete raw data in all three formats at no extra cost.
Get 10% Off with Code GENOME → Full raw data included · From €169 on flash sale