Airflow Omics · NASA OSDR

Overview

The flight response is real but largely specific to each experiment.

A flight effect shared by all 11 studies explains under 1 % of gene variance, against 7–11 % for study-specific flight effects. Within BRIC hardware, only the RNA-seq studies agree with each other. Neither CFD-predicted boundary-layer conductance nor carbon supply explains which studies respond, and the CFD shows why. Inside a real flight dish, gravity changes leaf gas exchange only about 1.4–1.8-fold, and sealed BRIC dishes run short of CO₂ on the ground as well as in orbit.

≤ 0.3 %

Shared flight response

Mean share of gene variance explained by a flight effect common to all studies (0.3 % pooled, under 0.01 % in RNA-seq alone).

r = 0.40

BRIC RNA-seq agreement

Median correlation of flight effects between BRIC RNA-seq studies. The BRIC microarray studies agree with none of them.

0 / 28

Pathways

Pathways with a flight effect or CFD dose-response across all studies once study structure is respected (one hit in a five-study subset is not replicated).

01The data

307 samples from 11 studies, in four hardware classes.

Horizontal bars of flight and ground-control sample counts for each of 11 OSDR studies, coloured by hardware class, with platform labels. — Flight and ground-control samples per study. APEX-TAGES (OSD-7) flight samples are grouped with BRIC; its ground controls were not sealed. EMCS (OSD-314) has no CFD scenario and is its own class. Source: `tables/reanalysis_2026-10/expression_metadata_with_cfd_v2.tsv`.

02A study-specific flight response

A flight response common to all studies explains under 1 % of gene variance.

Grouped bars of mean percent variance per component; study dominates, the common flight effect is 0.3 and under 0.01 percent, flight by study is 7.1 and 11.0 percent. — Variance partition with every factor as a random effect and study nested in hardware, for the pooled data and RNA-seq only. Source: REANALYSIS §12, `followups/variance_partition_summary_v2.tsv`.

03RNA-seq and microarray

BRIC RNA-seq studies share a flight signature; the BRIC microarray studies detect almost none.

Heatmap of correlations between flight effects in 17 study-by-organ units; the three BRIC RNA-seq units correlate around 0.4, microarray units correlate near zero with them. — Correlation of flight effects between study × organ units. The microarray studies are the earlier flights (BRIC, and APEX-TAGES for OSD-7), so platform and flight era cannot be separated. Source: REANALYSIS §13.

04No pathway-level flight or CFD effect

No pathway responds to flight across all studies once study structure is respected.

Dot plot of flight effects with 95 percent intervals for 28 pathways; no pathway reaches significance after multiple-testing correction. — Flight effect per pathway from a within-study model with cluster-robust errors, on pathway scores computed without zero-filling. Intervals are unadjusted; no pathway reaches FDR < 0.05, and the test detects only large effects. Source: REANALYSIS §10–§11.

05Inside the flight dish

The CFD explains why no boundary-layer dose-response appears.

Dumbbell chart of leaf boundary-layer conductance at 1 g and 0 g: open box 0.997 to 0.443, BRIC dish 0.33 to 0.185, CARA dish 0.263 to 0.186. — Leaf boundary-layer conductance at 1 g and 0 g, in an open box and inside the real BRIC and CARA dishes. Source: LunarLeaf-CFD T13 and T19.

Data & code

Every number and figure traces to a script and a table in the repository.

Component	Location
Findings and evidence	`reports/HEADLINE_2026-10.md`
Full re-analysis (§1–§13)	`tables/reanalysis_2026-10/`
Analysis code	`code/` (OSDR mining, CFD covariates, limma refits, figures)
Figures	`figures/2026-10/`
Manuscript	`manuscript.pdf`
CFD model	LunarLeaf-CFD