Format Atlas

Guides

Choosing Sequencing Files by Workflow

Raw reads, alignment inspection, analysis storage and sharing need different information. Manage references, indexes and preservation settings alongside formats.

Which format should you choose?

Choose by the role each file plays in your workflow.

  • Check reads and processing inputs
  • Validate alignments and indexes
  • Deliver a reproducible package

Check reads and processing inputs

Check FASTQ sequences, quality encoding, paired-end correspondence and read counts. Alignment needs a reference and tool settings. Converting an existing alignment representation is different from mapping reads to a reference.

Validate alignments and indexes

Inspect headers and tags in SAM, then store BAM or CRAM with compatible tools. For region queries, prepare coordinate sorting and the required BAI/CSI/CRAI index. Record CRAM reference retrieval or embedding and quality/tag preservation settings; compare before and after conversion.

Deliver a reproducible package

Include data, required indexes, reference identifiers, software versions, processing settings, preservation policy and checksums. Test the recipient's reader and agree on access and retention. Repository submissions have their own format and metadata requirements.

Formats to consider

FASTQ

Good for: Deliver raw reads and per-base qualities

Less suitable for: Store alignments with CIGAR and reference coordinates

BAM

Good for: Process alignments with compatible analysis tools

Less suitable for: Read contents using only a text editor

CRAM

Good for: Reduce storage with managed references and preservation settings

Less suitable for: Share without confirming reference resolution or quality retention

Before you export

  • Removing names alone does not necessarily anonymize human sequences. Confirm authorization and the destination's access controls before disclosure.
  • Keep originals until validation is complete. Matching checksums verify transfer integrity, not analytical correctness.

Related comparisons

FASTA vs FASTQ: sequences and quality retentionFASTA alone stores sequences; FASTQ stores sequences with per-base quality scores.SAM vs BAMSAM and BAM represent the same alignment information. SAM is TAB-delimited text; BAM is binary encoded in BGZF blocks.BAM vs CRAMBAM is a BGZF-compressed alignment representation. CRAM uses reference differences and other encodings; preservation depends on writer settings.

Guides

Handing off genomic annotations and signal tracksSeparate annotations from numerical signals and agree on the reference genome, coordinates, units and viewer. This is distinct from preserving raw reads or alignments.

References and specifications