read

A read is a single stretch of DNA sequence that comes out of a sequencing machine — one snippet of letters from one fragment. Since no instrument can read a whole genome in one go, the DNA is shattered into pieces and the machine produces a read for each piece, like tearing a book into scraps and transcribing each scrap.

Reads vary in length and quality depending on the technology. Short-read platforms produce reads of perhaps a hundred or a few hundred bases but enormous numbers of them; long-read platforms produce far fewer reads, each thousands of bases long. Each read also carries a per-base quality score reflecting how confident the machine is in each letter.

On their own, individual reads are just fragments; their power comes from quantity. Reading the same region with many overlapping reads — the depth of coverage — lets software cancel out random errors and reconstruct the true sequence, either by aligning the reads to a reference or by assembling them into longer pieces.