Classify bat sequences by regions and phase lengths

data/templates/bat_phase_analysis.py

@@ -1,4 +1,4 @@
-from batphase import detect_feeding_buzz_phases
+from batphase import detect_feeding_buzz_phases, classify
 from tabulate import build_table, print_table
 
 #: Pulse widths in ms
@@ -11,7 +11,10 @@
 DPRI = $DPRI
 
 #: Detect the feeding buzz phases
-_, regions, classification = detect_feeding_buzz_phases(WIDTHS, PRI, DPRI)
+_, regions = detect_feeding_buzz_phases(WIDTHS, PRI, DPRI)
+
+#: Classify the sequence
+classification = classify(regions)
 
 #: Build the phase table
 table = build_table(regions, ("Phase", "Start", "End", "Length"))

src/ecology/batphase.py

@@ -248,12 +248,52 @@ def detect_feeding_buzz_phases(widths, pri, dpri):
     for i in range(buzz_end + 1, n):
         phases[i] = EXIT
 
+    return phases, build_regions(phases)
+
+
+def classify(regions):
+    """
+    Classify a sequence based on its regions and their length
+
+    :param regions: List of tuples of regions
+    :return: Classification
+    """
+    total = 0
+    lengths = {}
+
+    # Collate the number of pulses by phase assignment
+    for phase, _, _, length in regions:
+        total += length
+        lengths[phase] = lengths.get(phase, 0) + length
+
+    # Extract the lengths of each phase assignment
+    search_pulses = lengths.get("SEARCH", 0)
+    approach_pulses = lengths.get("APPROACH", 0)
+    buzz_pulses = lengths.get("BUZZ", 0)
+    exit_pulses = lengths.get("EXIT", 0)
+
+    # Calculate the fraction of the total that are search pulses
+    search_frac = search_pulses / total if total else 0
+
+    # Determine which regions are present
+    has_buzz = buzz_pulses > 0
+    has_approach = approach_pulses > 0
+    has_exit = exit_pulses > 0
+
     # Classify the sequence
-    if BUZZ in phases:
+    if has_buzz and has_exit and buzz_pulses >= 6 and exit_pulses >= 5:
+        classification = "FEEDING PASS WITH RECOVERY EXIT"
+    elif has_buzz and buzz_pulses >= 6:
         classification = "FEEDING BUZZ"
-    elif APPROACH in phases:
-        classification = "APPROACH ONLY"
-    else:
+    elif has_buzz:
+        classification = "SHORT OR WEAK BUZZ"
+    elif has_approach and approach_pulses >= 5:
+        classification = "APPROACH WITHOUT BUZZ"
+    elif has_approach:
+        classification = "BRIEF APPROACH ONLY"
+    elif search_frac >= 0.85:
         classification = "SEARCH ONLY"
+    else:
+        classification = "MIXED OR UNCERTAIN PASS"
 
-    return phases, build_regions(phases), classification
+    return classification

src/examples/batbuzz.py

@@ -1,17 +1,20 @@
-from batphase import detect_feeding_buzz_phases
+from batphase import detect_feeding_buzz_phases, classify
 from tabulate import build_table, print_table
 
 #: Pulse widths in ms
-WIDTHS = (0.05111111111111111, 0.04693877551020406, 0.046485260770975034, 0.03356009070294785, 0.04643990929705216, 0.040362811791383235, 0.049024943310657654, 0.04435374149659865, 0.04825396825396833, 0.0386848072562358, 0.0338321995464852, 0.04022675736961445, 0.03573696145124727, 0.039682539682539764, 0.04240362811791387, 0.040181405895691746, 0.03959183673469391, 0.04367346938775518, 0.04349206349206347, 0.045941043083900235, 0.04589569160997731, 0.0401360544217686, 0.036190476190476106, 0.03396825396825398, 0.029251700680272164, 0.02285714285714291, 0.02408163265306107, 0.03224489795918384, 0.03396825396825376, 0.03519274376417236, 0.034739229024943086, 0.03795918367346918, 0.03428571428571425, 0.03609977324263047, 0.03256235827664433, 0.04548752834467118, 0.04081632653061229, 0.04099773242630356, 0.04004535147392252, 0.04040816326530594, 0.028888888888888964, 0.0394557823129249, 0.026848072562358105, 0.03859410430838972, 0.0439002267573696, 0.044625850340136, 0.04757369614512452, 0.04349206349206369, 0.04258503401360514, 0.03727891156462615)
+WIDTHS = (0.04240362811791387, 0.040181405895691746, 0.03959183673469391, 0.04367346938775518, 0.04349206349206347, 0.045941043083900235, 0.04589569160997731, 0.0401360544217686, 0.036190476190476106, 0.03396825396825398, 0.029251700680272164)
 
 #: Pulse Repetition Interval, PRI
-PRI = (0.08707482993197277, 0.09188208616780047, 0.0875283446712018, 0.08712018140589572, 0.09052154195011336, 0.09011337868480729, 0.09560090702947843, 0.07941043083900223, 0.08739229024943318, 0.09002267573696143, 0.08444444444444443, 0.081859410430839, 0.08802721088435383, 0.08263038548752832, 0.0840362811791382, 0.06390022675736962, 0.08589569160997734, 0.0734240362811791, 0.051337868480725746, 0.05764172335600892, 0.04335600907029491, 0.045578231292517035, 0.028843537414965814, 0.043809523809523965, 0.03428571428571425, 0.027936507936507926, 0.37918367346938786, 0.07800453514739214, 0.08585034013605464, 0.0790022675736961, 0.08331065759637157, 0.08195011337868507, 0.07818594104308385, 0.07183673469387752, 0.08258503401360562, 0.08145124716553287, 0.08485260770975023, 0.09065759637188231, 0.11800453514739218, 0.09024943310657596, 0.09283446712018151, 0.08947845804988663, 0.08916099773242614, 0.08580498866213171, 0.08825396825396803, 0.08916099773242658, 0.08829931972789096, 0.09814058956916138, 0.09247165532879809, None)
+PRI = (0.0840362811791382, 0.06390022675736962, 0.08589569160997734, 0.0734240362811791, 0.051337868480725746, 0.05764172335600892, 0.04335600907029491, 0.045578231292517035, 0.028843537414965814, 0.043809523809523965, None)
 
 #: Delta-PRI, DPRI
-DPRI = (None, 0.0048072562358277005, -0.004353741496598673, -0.00040816326530607183, 0.0034013605442176353, -0.00040816326530607183, 0.005487528344671144, -0.0161904761904762, 0.007981859410430947, 0.002630385487528253, -0.005578231292517, -0.002585034013605436, 0.006167800453514838, -0.005396825396825511, 0.0014058956916098708, -0.02013605442176858, 0.021995464852607727, -0.012471655328798237, -0.02208616780045336, 0.006303854875283177, -0.014285714285714013, 0.0022222222222221255, -0.01673469387755122, 0.01496598639455815, -0.009523809523809712, -0.006349206349206327, 0.35124716553287993, -0.3011791383219957, 0.007845804988662497, -0.0068480725623585315, 0.004308390022675468, -0.001360544217686499, -0.0037641723356012236, -0.006349206349206327, 0.010748299319728094, -0.0011337868480727487, 0.003401360544217358, 0.005804988662132082, 0.027346938775509866, -0.027755102040816215, 0.002585034013605547, -0.003356009070294874, -0.00031746031746049397, -0.00335600907029443, 0.00244897959183632, 0.0009070294784585542, -0.0008616780045356265, 0.009841269841270428, -0.005668934240363299, None)
+DPRI = (None, -0.02013605442176858, 0.021995464852607727, -0.012471655328798237, -0.02208616780045336, 0.006303854875283177, -0.014285714285714013, 0.0022222222222221255, -0.01673469387755122, 0.01496598639455815, None)
 
 #: Detect the feeding buzz phases
-_, regions, classification = detect_feeding_buzz_phases(WIDTHS, PRI, DPRI)
+_, regions = detect_feeding_buzz_phases(WIDTHS, PRI, DPRI)
+
+#: Classify the sequence
+classification = classify(regions)
 
 #: Build the phase table
 table = build_table(regions, ("Phase", "Start", "End", "Length"))

src/support/extract_pulses.py

@@ -52,37 +52,70 @@
 import os
 import argparse
 from pathlib import Path
+from datetime import datetime
 
 DEFAULT_TEMPLATE = Path(__file__).parent.parent.parent / "data" / "templates" / "bat_pulse_chart.py"
 
 
-def load_pulse_json(input_file_path: str | Path) -> dict:
+def print_message(message: str) -> None:
+    """
+    Show a timestamped message
+
+    :param message: Message text
+    """
+    timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    print(f"{timestamp} : {message}")
+
+
+def load_pulse_json(
+        input_file_path: str | Path,
+        start_index: int = 1,
+        end_index: int = None
+) -> dict:
     """
     Load the consensus parameter set from the seasonal modelling consensus JSON
     file for the species
 
     :param input_file_path: JSON file path
     :return: Tuple of the species namd and the filtered list of parameters
     """
+    print_message(f"Loading {input_file_path}")
     with open(input_file_path, "r", encoding="utf-8") as f:
         data = json.load(f)
 
     # Extract the original input file and extract it's name as the source
     source_file = data.get("input", data.get("input_file", None))
     source = Path(source_file).name
+    print_message(f"Source file name {source}")
 
     # Extract the analysis mode
     mode = data["analysis_mode"]
+    print_message(f"Detected mode {mode}")
+
+    # Extract only the required range of pulses
+    pulses = data["pulses"]
+    max_pulse_index = len(pulses)
+    print_message(f"Detected {max_pulse_index} pulses")
+
+    if start_index is None or start_index <= 0:
+        start_index = 1
+
+    if end_index is None or end_index > max_pulse_index:
+        end_index = max_pulse_index
+
+    pulses = pulses[start_index - 1:end_index]
+    print_message(f"{len(pulses)} pulses remain after extracting the slice from {start_index} to {end_index}")
 
     # Extract the pulses. Use the "real" start time, end time and peak timing fiels.
     # For heterodyne recordings, these will be the same as the "non-real" versions
     # but for time expansion the "non-real" versions reflect the expanded timings
     timings = []
-    for pulse in data["pulses"]:
+    for pulse in pulses:
         timings.append(pulse["real_start_time_s"])
         timings.append(pulse["real_end_time_s"])
         timings.append(pulse["real_peak_time_s"])
 
+    print_message(f"Extracted {len(timings) // 3} sets of timing values")
     return source, mode, tuple(timings)
 
 
@@ -93,6 +126,7 @@ def read_template(template_file_path: str | Path) -> str:
     :param template_file_path: Path to the template
     :return: Contents of the template
     """
+    print_message(f"Loading template {template_file_path}")
     with open(template_file_path, "r", encoding="utf-8") as f:
         return f.read()
 
@@ -116,6 +150,7 @@ def write_analysis_script(output_path: str | Path, script: str):
     :param output_path: Path to the folder where the script is to be written
     :param script: File contents
     """
+    print_message(f"Writing launcher {output_path}")
     with open(output_path, "w", encoding="utf-8") as f:
         return f.write(script)
 
@@ -128,11 +163,13 @@ def main():
     parser.add_argument("-i", "--input", required=True, help="Path to the input JSON file")
     parser.add_argument("-t", "--template", default=DEFAULT_TEMPLATE,
                         help="Template used to build the bat call analysis script")
+    parser.add_argument("-si", "--start-index", type=int, default=1, help="Index of first pulse to include")
+    parser.add_argument("-ei", "--end-index", type=int, default=None, help="Index of last pulse to include")
     parser.add_argument("-o", "--output", required=True, help="Path to the output script")
     args = parser.parse_args()
 
     # Load the data and extract the pulse information into a tuple
-    _, _, pulses = load_pulse_json(args.input)
+    _, _, pulses = load_pulse_json(args.input, args.start_index, args.end_index)
 
     # Load the template and generate the script content
     template = read_template(args.template)

src/support/extract_timings.py

@@ -45,6 +45,9 @@
 
     scripts/extract-timings.sh -i data/spectrogram/BD-A-99-001-analysis.json -o src/examples/batbuzz.py
 
+If only part of the sequence is to be included, the --start-index and --end-index arguments can
+be used to specify the 1-based indices for the first and last pulse to include
+
 The resulting script can then be minified and transferred to the calculator along with
 the bat phase analysis library module and then run to generate the pulse characteristics.
 
@@ -114,11 +117,13 @@ def main():
     parser.add_argument("-i", "--input", required=True, help="Path to the input JSON file")
     parser.add_argument("-t", "--template", default=DEFAULT_TEMPLATE,
                         help="Template used to build the bat phase analysis script")
+    parser.add_argument("-si", "--start-index", type=int, default=1, help="Index of first pulse to include")
+    parser.add_argument("-ei", "--end-index", type=int, default=None, help="Index of last pulse to include")
     parser.add_argument("-o", "--output", required=True, help="Path to the output script")
     args = parser.parse_args()
 
     # Load the pulse data from the JSON file and generate the pulse timing information
-    _, _, pulses = load_pulse_json(args.input)
+    _, _, pulses = load_pulse_json(args.input, args.start_index, args.end_index)
     widths, pri, _, dpri = analyse_pulse_timings(pulses)
 
     # Load the template and generate the script content

src/support/minimiser.conf

@@ -38,7 +38,7 @@
         "batphase.py": {
             "aggressive": true,
             "preserve": [
-                "classify_sequence",
+                "classify",
                 "detect_feeding_buzz_phases"
             ]
         },