XenoPanel

What it does

Targeted panels are cheap and sensitive at low input, but they raise two problems a general RNA-seq pipeline does not solve: cross-species signal in xenografts, and telling apart which genes change from why they change. The toolkit handles both, alongside standard differential expression and several transcriptome-state read-outs. Inputs, modules, thresholds and contrasts are set in one config file; nothing is hard-wired to a single experiment.

Four modules

Each can be switched on or off in the config.

Cross-reactivity

Estimate, per feature, the fraction of signal from a second (host) species using a dilution series, and build an exclusion mask.

QC + differential expression

filterByExpr, TMM normalization, per-sample quality control, and edgeR quasi-likelihood F-tests with TREAT across any contrasts you define.

Decomposition

Split each gene's change into heritable and environmental components via an exact additive identity over three states, with optional site attribution.

Entropy + pathway activity

Transcriptome diversity, SCENT signalling entropy and CCAT, signature scoring, PROGENy pathway activity, and CollecTRI transcription-factor activity.

Example outputs

From the bundled synthetic example: host-signal masking, differential expression, the heritable/environmental decomposition, and network entropy by group.

Quick start

R 4.3+ with yaml and edgeR; other packages only for the modules that use them.

# generate the synthetic example data
Rscript examples/make_example_data.R

# run the whole pipeline
Rscript run.R config.example.yaml

# confirm it recovers the planted truth
Rscript tests/test_recovery.R

To run on your own data, copy config.example.yaml, edit the paths, contrasts and thresholds, and run Rscript run.R your_config.yaml. See the repository for the full guide, input format and requirements.