DataFrame Inspection

FRUST workflow outputs are usually inspected as pandas DataFrames. The DataFrame plotting helpers convert rows and coordinate columns into molecular views.

Read FRUST Output

import pandas as pd

df = pd.read_parquet("runs/example/init.hess.optts.freq.solv.parquet")

Filter failures before ranking

A failed row may still contain metadata or partial coordinates. Filter on the final *-NT column before comparing energies or selecting examples to visualize.

nt_cols = [col for col in df.columns if col.endswith("-NT")]
df_ok = df[df[nt_cols[-1]]]

Plot Molecules From Rows

from frust.vis import plot_mols

plot_mols(df_ok)

Representative output:

The helper searches coordinate-like columns such as coords, *-oc, and *-opt_coords, converts atoms and coordinates to molecules, and shows them in a grid.

Inspect one row

from frust.vis import plot_row

plot_row(df_ok, row_index=0)

Representative output:

Inspect a substrate or reactive position

from frust.vis import plot_lig, plot_rpos

plot_lig(df_ok, "anisole")
plot_rpos(df_ok, 4)

Representative output:

Select Coordinate Columns

Use include_coords, exclude_coords, or coord_indices when a workflow has many optimization stages.

plot_mols(
    df_ok,
    include_coords=["OptTS", "Freq"],
    cell_size=(450, 450),
)

Representative output is still an interactive molecule grid like the examples above, but restricted to the selected coordinate stages.

Start broad, then narrow

First call plot_mols(df_ok.head()) to see which coordinate columns are present. Then narrow to the stages that matter for your inspection.