What files Aimosti accepts

A VCF lists only where you differ from the reference. It does not say which positions were read. So when a disease variant is not in the file, two explanations fit: you do not carry it, or the sequencer never covered that spot. A gVCF records the covered regions, which separates those two cases. Both files give the full report. The gVCF lets us say "examined and clear" instead of "not found".

A genotyping chip reads a few hundred thousand chosen positions. A sequenced genome reads billions. From a chip you get traits, ancestry, and pharmacogenomics with every defining position shown as read, present but unresolvable, or absent from your chip. You also get a few specific findings that sit on standard chips, such as Factor V Leiden and Prothrombin G20210A, reported for that variant alone. You do not get the full clinical-findings panel or carrier status. Those need a sequenced genome, because a chip never reads most of those positions. We say so here rather than after you pay.

GRCh38 and GRCh37. Some providers, Dante among them, still deliver GRCh37. On GRCh37 the 29-gene pharmacogenomic panel runs in full. The wider clinical coverage map and the recurrent-deletion screen are GRCh38-only; on a GRCh37 file we report them as not measured. We do not estimate them. Contig names match across builds, so a file read against the wrong build would produce numbers at the wrong coordinates instead of an error. That is why we detect the build from your file and act on what we find.

Reads are the evidence a variant file was summarised from. Some genes cannot be called from a summary at all. CYP2D6 is the standard example: duplications, deletions and a near-identical pseudogene defeat a variant file, and CYP2D6 drives the dosing of common antidepressants, painkillers and tamoxifen. The €109 Deep Read runs a 29-gene panel from your reads and measures how well each gene was covered, so you can see which answers are solid.

No. Gzipped, bgzipped and zipped files all work. Send the file the way your provider sent it. Converting it yourself is the most common way to break it.

Three things. Unaligned reads (FASTQ): align them first, or ask your provider for the variant file. Multi-sample and trio VCFs: upload one person per file. Sites-only VCFs that carry no genotypes: the file lists positions but not what you carry. We reject these at upload and tell you which one it was, rather than analysing a file we cannot read correctly.