1. Basic usage¶

path=os.path.expanduser("~/FunEcoLab_IBFG Dropbox/Noise/Datasets/")

L = ep.landscape_from_file(
    os.path.join(path, "Complete_landscape_Sabela.csv"),
)

display(L.to_dataframe)

# --- Demo DataFrame ---
df = pd.DataFrame({
    "g0": [0, 0, 0, 1, 1, 1, 0, 1],
    "g1": [0, 0, 1, 0, 1, 1, 1, 0],
    "g2": [0, 1, 0, 0, 0, 1, 1, 1],
    "rep_1": [1.00, 1.10, 1.20, 1.30, np.nan, 1.50, 1.60, 1.70],
    "rep_2": [1.05, 1.12, np.nan, 1.28, 1.38, 1.48, 1.58, 1.68],
    "rep_3": [0.95, 1.08, 1.22, 1.33, 1.41, np.nan, 1.61, 1.69],
})

header("Demo DataFrame (first rows)")
display(df)

# Show basic properties of the dataset
header("Landscape summary")
print(f"N (dimensions):      {L.N}")       # Number of binary variables
print(f"R (replicates):      {L.R}")       # Number of experimental replicates
print(f"order:               {L.order}")   # State ordering ('lex' = lexicographic)
print(f"M (observed states): {L.M}")       # Number of observed states (rows)
print(f"Feature names:       {L.feature_names}")  # List of feature names
print(L.states)                            # Binary matrix (M × N)

# Compute mean value per state, ignoring NaNs
mean = L.mean_over_replicates()
mean_df = pd.DataFrame({
    "State index": range(L.M),
    "Mean across replicates": mean
})
display(mean_df)

# --- Select a subset of replicate columns (keeps all states) ---
L_subset = L.select_replicates([0, 2])  # Select only replicates 0 and 2

# Convert to DataFrame for inspection
subset_df = pd.DataFrame(
    np.hstack([L_subset.states, L_subset.values]),
    columns=[f"s{i}" for i in range(L_subset.N)] +
            [f"rep_{j}" for j in range(L_subset.R)]
)

print("Subset with replicate columns 0 and 2:")
display(subset_df)

# --- Select a subset of states (rows) ---
# Example 1: Select by explicit indices
L_sel_idx = L.select_states([0, 3, 5])

# Example 2: Select only rows without NaNs in the first replicate
mask_valid = ~np.isnan(L.values[:, 0])
L_sel_mask = L.select_states(mask_valid)

# Convert one of them to DataFrame for display
sel_df = pd.DataFrame(
    np.hstack([L_sel_idx.states, L_sel_idx.values]),
    columns=[f"s{i}" for i in range(L_sel_idx.N)] +
            [f"rep_{j}" for j in range(L_sel_idx.R)]
)

print("Subset with selected state indices [0, 3, 5]:")
display(sel_df)

# --- Retrieve replicate values for a specific state ---
# Example 1: Pass the binary state as a list
vals = L.get_values([1, 0, 1])
print("Replicate values for state [1, 0, 1]:")
print(vals)

# Example 2: Retrieve only some replicates
vals_sub = L.get_values([1, 0, 1], replicates=[0, 1])
print("\nOnly first two replicates:")
print(vals_sub)

# Example 3: Return a labeled DataFrame
vals_df = L.get_values([1, 0, 1], as_dataframe=True)
display(vals_df)

# Example 4: Retrieve using integer encoding (binary 101 = 5)
print("\nBy integer encoding (5):")
print(L.get_values(5))

###############################################################
# 0 Create new demo dataset with three types of missing data  #
###############################################################

#State [1 1 1] has been removed
df_missing = pd.DataFrame({
    "g0": [0, 0, 0, 1, 1, 0, 1],
    "g1": [0, 0, 1, 0, 1, 1, 0],
    "g2": [0, 1, 0, 0, 0, 1, 1], 
    "rep1": [1.00, 1.10, 1.20, 1.30, np.nan, 1.60, 1.70],
    "rep2": [1.05, 1.12, np.nan, 1.28, np.nan, 1.58, 1.68],
    "rep3": [0.95, 1.08, 1.22, 1.33, np.nan,  1.61, 1.69],
})


header("Demo DataFrame with missing data")
display(df_missing)

#Load as a Landscape object
L_missing = ep.Landscape.from_dataframe(df_missing) 

#################################################
# 1 Strict filtering: remove rows with ANY NaN  #
#################################################

L_complete = L_missing.drop_rows_with_any_nan()
clean_df = pd.DataFrame(
    np.hstack([L_complete.states, L_complete.values]),
    columns=[f"g{i}" for i in range(L_complete.N)] +
            [f"rep_{j}" for j in range(L_complete.R)]
)


print("1-STRICT FILTERING: no missing replicates")
display(clean_df)

#####################################################
# 2 Permissive filtering: remove rows with ALL NaNs #
#####################################################

# Introduce one row that is completely NaN in all replicates
L_partial = L_missing.drop_rows_with_all_nan()
partial_df = pd.DataFrame(
    np.hstack([L_partial.states, L_partial.values]),
    columns=[f"g{i}" for i in range(L_partial.N)] +
            [f"rep_{j}" for j in range(L_partial.R)]
)

print("2-PERMISSIVE FILTERING: REMOVE ROWS WITH ALL NaNS")
display(partial_df)

###########################################################
# 3 Missing states: configurations not present in dataset #
###########################################################

print("3-MISSING STATES: CONFIGURATIONS NOT PRESENT IN DATASET")

missing = L_missing.missing_states()
print(f"\nMissing {len(missing)} states out of {2**L_missing.N} total:")
display(pd.DataFrame(missing, columns=[f"g{i}" for i in range(L_missing.N)]))