Log in to add personal notes on this page.
Long-read sequencing reads single DNA molecules thousands to millions of bases long, compared with the 150 to 300 bases of a short-read NGS read. A read that long can span a repeat expansion, both breakpoints of a rearrangement, or a gene together with its nearly identical pseudogene. Because it reads native DNA without PCR, it also detects methylation. It is used mostly when short-read testing is negative or for regions short reads cannot resolve, and a growing number of clinical labs offer it as a genome test.
- PacBio HiFi: DNA is closed into a circle, and a polymerase copies it around and around, so the instrument reads the same molecule many times. The consensus of those passes gives reads of about 15 to 20 kb with a median accuracy of about 99.9% per base.
- Oxford Nanopore: A single DNA strand is pulled through a protein pore, and each stretch of bases changes the electrical current in its own way. Reads are routinely tens of kilobases and can exceed a million bases. Per-base accuracy has risen to about 99% with current chemistry; the remaining errors cluster in runs of a single base (homopolymers). Instruments range from benchtop machines to a handheld device.
- No amplification: Both platforms read native DNA. That avoids PCR bias in GC-rich regions and lets base modifications, mainly 5-methylcytosine, be read directly from the same molecule.
- Phasing: A read that covers two variants shows whether they lie on the same chromosome copy (in cis) or on opposite copies (in trans). Long reads can often settle this without parental samples.
- High-molecular-weight DNA: Long reads need long, intact DNA, usually extracted from fresh blood or cells. DNA prepared for routine short-read testing is often too fragmented.
- Repeat expansions: One read spans the whole repeat and gives its size, interruptions, and methylation in one test. This helps at FMR1, C9orf72, and DMPK, and with the biallelic RFC1 expansion of CANVAS (cerebellar ataxia, neuropathy, and vestibular areflexia syndrome), which short-read genome analysis often misses. See Repeat expansion testing.
- Genes with pseudogenes or close copies: When two sequences are nearly identical, a short read cannot tell which copy it came from; a long read can. This helps at SMN1 and SMN2 in spinal muscular atrophy, CYP21A2 and its pseudogene CYP21A1P in congenital adrenal hyperplasia, PKD1 in autosomal dominant polycystic kidney disease, GBA1 in Gaucher disease, the HBA1 and HBA2 deletions of alpha thalassemia, and CYP2D6 in pharmacogenomic testing.
- Structural variants: Long reads cross breakpoints, so they detect balanced translocations, inversions, and insertions (including mobile element insertions). Microarray cannot see a balanced change at all, and short-read genome sequencing misses one when a breakpoint falls in repetitive sequence. The F8 intron 22 inversion, found in about 45% of severe hemophilia A, has its breakpoints inside a large repeated segment, which is why short reads cannot see it. See Structural variants.
- Two variants in a recessive gene: When a proband carries two variants in the same recessive gene and parental samples are not available, long reads can show whether they are in trans, which is needed before the result explains the condition.
- Methylation from the same run: The same data show imprinting defects and FMR1 promoter methylation with no bisulfite step. See Methylation testing.
- After a negative short-read genome: Long-read genome sequencing finds a diagnosis in a minority of unsolved cases, mostly structural variants, repeat expansions, and variants in genes that short reads map poorly.
- Cost and availability: A long-read genome still costs more than a short-read genome, and fewer labs offer it clinically.
- Sample requirements: It needs high-molecular-weight DNA, which many stored or degraded samples cannot provide.
- Error profile: Nanopore reads still make more small insertion and deletion errors in homopolymers than short reads do, which matters when calling small indels.
- Interpretation: Long reads find many more structural variants and repeat alleles per genome, and population databases for these are smaller than those for single-base variants, so more findings are of uncertain significance.
- Validation: Clinical standards and validated analysis pipelines are still catching up with the technology.
- Long reads are the answer when the suspected change sits in a repeat, a rearrangement, or a gene with a pseudogene, because those are the three places short reads are weakest.
- Long reads report methylation from the same run, with no bisulfite step.
- HiFi reads one circular molecule many times for accuracy; nanopore threads one strand through a pore for length.