Reliability of copy-specific RNA quantification in seven tandem gene pairs of Lumnitzera littorea under salt treatment

Combining the expression of similar gene copies can conceal changes in their relative RNA contributions, but distinguishing the copies in RNA-seq data is difficult. We examined this problem using public Lumnitzera littorea genome and RNA-seq data. Each of two datasets contained three control libraries and three libraries from plants treated with 400 mM NaCl for three days. Leaves were the primary dataset. Roots were analyzed separately because inconsistent reference-species metadata made their interpretation conditional. Seven of 97 tandem pairs met the structural and simulated read-discrimination criteria. Four leaf pairs and two root pairs also had sufficient counts for compositional analysis. A second reference recovered all 14 coding models and their A/B assignments. Uniquely assigned counts and Salmon agreed on the direction of change in all six count-eligible pair–study combinations. That direction persisted when RNA was requantified against the second genome and transferred annotation; the largest per-library difference in copy fraction was 0.88 percentage points. In the leaf cystatin-like pair, the fitted A fraction fell from 22.94% to 8.16%, while normalized pair-total expression changed by approximately 3.5% in a descriptive comparison. Within 1 kb upstream of the coding boundaries, the transposable-element (TE) label groups examined covered 0% of A and 41.8% of B. The DNA and RNA measurements were neither matched within individuals nor linked by functional tests. The minimum attainable two-sided label-check p-value was 0.10. Null simulations also supported treating the fraction estimates as exploratory. Resolving the copies revealed RNA patterns that persisted across the measurement choices tested. Their duplication origin and any role in salt tolerance remain unresolved. This preprint has not been peer reviewed. Supporting data and analysis code: 10.5281/zenodo.23967900.