SniffCell preprint: long DNA reads identify cell types and reveal recurrent repeat expansions in neurons
SniffCell preprint: long DNA reads identify cell types and reveal recurrent repeat expansions in neurons
On September 4, the authors of a preprint described SniffCell. The method uses methylation, chemical marks on DNA, to assign a long DNA molecule from mixed tissue to a cell type. The authors checked these assignments against physically sorted cells from blood and postmortem brain tissue, then identified recurrent tandem repeat expansions associated with neurons in the brain.
A frontal cortex sample contains a mixture of neurons and oligodendrocytes, cells that form an insulating sheath around nerve fibers. A somatic variant is a DNA change that arises in cells of the body during development or later in life. When mixed tissue is analyzed, the variant appears as a signal from the sample as a whole, obscuring its cellular origin.
A long read captures an extended stretch of a single DNA molecule, retaining both its sequence and its methylation marks. The authors used a 2023 methylation atlas covering 39 cell groups from 205 healthy samples. SniffCell extended this atlas for eight tissues: it compares the marks on each read with the cell-type map, then searches for somatic variants within the resulting groups.
The authors checked the assignments against 16 physically sorted fractions from blood and brain tissue. Mean accuracy was 80,38%, the proportion of reads assigned to the expected cell type or a closely related group. In blood, SniffCell detected the expected T-cell and B-cell receptor rearrangements and assigned one somatic tandem repeat to T-cells.
In four frontal cortex samples, 83 of 92 DNA regions with cell-type-specific tandem repeats were confirmed in at least one comparison with a sorted fraction. The authors then examined three postmortem brain cohorts comprising 172 donors. The three-letter DNA sequence GAA was more common among recurrent expansions associated with neurons.
Blood from the same donor provides the inherited repeat length as a reference for comparison with brain cells. For ten donors with recurrent changes in neurons and oligodendrocytes, the median proportion of events for which the expanded variant was also found in blood was 29,2% for neuronal events and 66,7% for oligodendrocyte events. A review of genomic mosaicism discussed how to connect a somatic mutation to tissue function. SniffCell makes it possible to examine an earlier step in that connection: identifying the cell population in which repeat length differs.