4.6.3 Gene correlation¶

Function¶

rpf_Corr evaluates sample reproducibility from a merged density table. It calculates Pearson correlation matrices at both gene level and RPF-position level, then draws heatmaps for all-frame density and each individual reading frame.

Input files¶

Input	Required	Description
Merged density table	Yes	RPF or RNA density table generated by `rpf_Merge`, usually `RIBO_merged.txt` or `RNA_merged.txt`. The table should contain per-sample frame columns such as `<sample>_f0`, `<sample>_f1`, and `<sample>_f2`.

Parameters¶

Parameter	Required	Description
`-r`	Yes	Input RPF or RNA density table in TXT format.
`-o`	Yes	Output prefix.

Output files¶

Assuming -o RIBO, the command writes:

Output	Description
`RIBO_gene_corr_frame.txt`	Gene-level Pearson correlation matrix using all-frame density.
`RIBO_gene_corr_f0.txt`	Gene-level Pearson correlation matrix using frame 0 density.
`RIBO_gene_corr_f1.txt`	Gene-level Pearson correlation matrix using frame 1 density.
`RIBO_gene_corr_f2.txt`	Gene-level Pearson correlation matrix using frame 2 density.
`RIBO_rpf_corr_frame.txt`	RPF-position-level Pearson correlation matrix using all-frame density.
`RIBO_rpf_corr_f0.txt`	RPF-position-level Pearson correlation matrix using frame 0 density.
`RIBO_rpf_corr_f1.txt`	RPF-position-level Pearson correlation matrix using frame 1 density.
`RIBO_rpf_corr_f2.txt`	RPF-position-level Pearson correlation matrix using frame 2 density.
`RIBO_gene_correlation_plot.pdf` / `.png`	Heatmap panel of gene-level correlation matrices.
`RIBO_rpf_correlation_plot.pdf` / `.png`	Heatmap panel of RPF-position-level correlation matrices.

Example¶

rpf_Corr \
    -r ../05.merge/RIBO_merged.txt \
    -o RIBO \
    &> RIBO.corr.log

Notes¶

Gene-level correlation is calculated after summing CDS density by gene.
RPF-position-level correlation is calculated from nucleotide/codon-position density across the merged table.
Replicates should generally cluster together. Low correlation may indicate sequencing depth differences, library quality problems, annotation mismatch, or biological divergence.
Correlation should be interpreted together with PCA, metagene periodicity, and coverage profiles.