Python API

scanRBP's scoring logic is available directly from Python, for embedding into other tools — this is how the expressRNA scanRBP analysis calls it under the hood.

Scoring a sequence

import scanRBP
df = scanRBP.scan("AAAGCGGCGACTTATTATATCCCCATATATTATATCTTCTTCTCTTATATATAAACCAGAGATAGATGTGTGTGGTGG")

scan(seq) returns a pandas.DataFrame indexed by scan_id (one row per RBP in the database), with one column per position in seq — the same log-odds score matrix the command line writes to .tab.gz.

Note

scan() always floors negative scores to 0 (equivalent to the CLI's -nonzero flag) — unlike the plain CLI invocation, which keeps negative scores by default. Keep this in mind if you're comparing Python API output against a CLI run made without -nonzero.

df.loc["TARDBP.K562.00"]       # score vector for one RBP across the sequence
df.sum(axis=1).sort_values()   # total binding score per RBP, ranked

Inspecting the database

import scanRBP

# scan_id -> {protein, tissue, source, cluster, pwm_path, aliases, description, ...}
scanRBP.database.proteins["TARDBP.K562.00"]

# scan_id -> biopython PSSM (log-odds matrix) used by scan()
scanRBP.pwm.pssm["TARDBP.K562.00"]

Both scanRBP.database.proteins and scanRBP.pwm.pssm are populated once, automatically, when you import scanRBP (triggering the same PWM database download described in Motif database & search if it isn't present yet) — there's no separate init call needed.

Configuration

import scanRBP
scanRBP.config.data_folder   # currently configured data folder (default ~/scanRBP_data)
scanRBP.config.init("/path/to/data_folder")  # change it, same as `scanRBP config /path` on the CLI