Finding. Espinoza, Silva and Pellegrini run the same human PBMCs and CD8 TEMRA cells on 10X Genomics and Parse. The kits do not tell the same story. Mitochondrial and ribosomal capture differ. 10X likes short genes. Parse likes long ones. On TEMRAs, Parse underplays GNLY, PRF1, and GZMB relative to 10X. Platform, not biology, can decide what a killer cell looks like and how “mitochondrial” a droplet is.
Why this is on Mito.news
Percent-mitochondrial reads are a default QC and sometimes a biology claim. If the kit moves that number, every immunometabolism atlas that mixed vendors is a little fictional.
Score 57. Methods warning. High confidence in the comparison as stated. Skip if you wanted a new organelle mechanism.
What to do with it Put a platform covariate in any 10X-plus-Parse merge. Do not call a TEMRA “low cytotoxicity” on Parse without a 10X check.