masster 0.3.13__py3-none-any.whl → 0.3.15__py3-none-any.whl

masster/study/processing.py

@@ -683,6 +683,15 @@ def merge(self, **kwargs):
         min_samples = 1
     if min_samples < 1:
         min_samples = int(min_samples * len(self.samples_df))
+    
+    # Validate that min_samples doesn't exceed the number of samples
+    if min_samples > len(self.samples_df):
+        self.logger.warning(
+            f"min_samples ({min_samples}) exceeds the number of samples ({len(self.samples_df)}). "
+            f"Setting min_samples to {len(self.samples_df)}."
+        )
+        min_samples = len(self.samples_df)
+    
     # filter out consensus features with less than min_samples features
     l1 = len(self.consensus_df)
     self.consensus_df = self.consensus_df.filter(

masster/study/study.py

@@ -59,6 +59,7 @@ from masster.study.h5 import _save_study5_compressed
 from masster.study.helpers import _get_consensus_uids
 from masster.study.helpers import _get_feature_uids
 from masster.study.helpers import _get_sample_uids
+from masster.study.helpers import _ensure_features_df_schema_order
 from masster.study.helpers import compress
 from masster.study.helpers import compress_features
 from masster.study.helpers import compress_ms2
@@ -77,14 +78,19 @@ from masster.study.helpers import get_gaps_stats
 from masster.study.helpers import align_reset
 from masster.study.helpers import set_folder
 from masster.study.helpers import set_source
-from masster.study.helpers import name_replace
-from masster.study.helpers import name_reset
+from masster.study.helpers import sample_color
+from masster.study.helpers import sample_color_reset
+from masster.study.helpers import sample_name_replace
+from masster.study.helpers import sample_name_reset
+from masster.study.helpers import samples_select
+from masster.study.helpers import samples_delete
 from masster.study.helpers import features_select
 from masster.study.helpers import features_filter
 from masster.study.helpers import features_delete
 from masster.study.helpers import consensus_select
 from masster.study.helpers import consensus_filter
 from masster.study.helpers import consensus_delete
+from masster.study.helpers import migrate_map_id_to_index
 from masster.study.load import add
 from masster.study.load import add_sample
 from masster.study.load import fill_single
@@ -276,22 +282,30 @@ class Study:
                 "sample_name": [],
                 "sample_path": [],
                 "sample_type": [],
-                "size": [],
                 "map_id": [],
-                "file_source": [],
-                "ms1": [],
-                "ms2": [],
+                "sample_source": [],
+                "sample_color": [],
+                "sample_group": [],
+                "sample_batch": [],
+                "sample_sequence": [],
+                "num_features": [],
+                "num_ms1": [],
+                "num_ms2": [],
             },
             schema={
                 "sample_uid": pl.Int64,
                 "sample_name": pl.Utf8,
                 "sample_path": pl.Utf8,
                 "sample_type": pl.Utf8,
-                "size": pl.Int64,
-                "map_id": pl.Utf8,
-                "file_source": pl.Utf8,
-                "ms1": pl.Int64,
-                "ms2": pl.Int64,
+                "map_id": pl.Int64,
+                "sample_source": pl.Utf8,
+                "sample_color": pl.Utf8,
+                "sample_group": pl.Utf8,
+                "sample_batch": pl.Int64,
+                "sample_sequence": pl.Int64,
+                "num_features": pl.Int64,
+                "num_ms1": pl.Int64,
+                "num_ms2": pl.Int64,
             },
         )
         self.features_maps = []
@@ -362,8 +376,12 @@ class Study:
     fill_reset = fill_reset
     align_reset = align_reset
     set_source = set_source
-    name_replace = name_replace
-    name_reset = name_reset
+    sample_color = sample_color
+    sample_color_reset = sample_color_reset
+    name_replace = sample_name_replace
+    name_reset = sample_name_reset
+    samples_select = samples_select
+    samples_delete = samples_delete
     features_select = features_select
     features_filter = features_filter
     features_delete = features_delete
@@ -387,6 +405,7 @@ class Study:
     _get_consensus_uids = _get_consensus_uids
     _get_feature_uids = _get_feature_uids
     _get_sample_uids = _get_sample_uids
+    _ensure_features_df_schema_order = _ensure_features_df_schema_order
     get_consensus_matrix = get_consensus_matrix
     get_gaps_matrix = get_gaps_matrix
     get_gaps_stats = get_gaps_stats
@@ -685,31 +704,6 @@ class Study:
 
         print(summary)
 
-    def _ensure_features_df_schema_order(self):
-        """
-        Ensure features_df columns are ordered according to study5_schema.json.
-        
-        This method should be called after operations that might scramble the column order.
-        """
-        if self.features_df is None or self.features_df.is_empty():
-            return
-        
-        try:
-            import os
-            import json
-            from masster.study.h5 import _reorder_columns_by_schema
-            
-            # Load schema
-            schema_path = os.path.join(os.path.dirname(__file__), "study5_schema.json")
-            with open(schema_path, 'r') as f:
-                schema = json.load(f)
-            
-            # Reorder columns to match schema
-            self.features_df = _reorder_columns_by_schema(self.features_df, schema, 'features_df')
-            
-        except Exception as e:
-            self.logger.warning(f"Failed to reorder features_df columns: {e}")
-
 
 if __name__ == "__main__":
     # This block is executed when the script is run directly

masster/study/study5_schema.json

@@ -225,9 +225,9 @@
         "dtype": "pl.Int64"
       },
       "map_id": {
-        "dtype": "pl.Utf8"
+        "dtype": "pl.Int64"
       },
-      "file_source": {
+      "sample_source": {
         "dtype": "pl.Utf8"
       },
       "sample_name": {
@@ -239,13 +239,22 @@
       "sample_type": {
         "dtype": "pl.Utf8"
       },
-      "size": {
+      "sample_group": {
+        "dtype": "pl.Utf8"
+      },
+      "sample_batch": {
+        "dtype": "pl.Int64"
+      },
+      "sample_sequence": {
+        "dtype": "pl.Int64"
+      },
+      "num_features": {
         "dtype": "pl.Int64"
       },
-      "ms1": {
+      "num_ms1": {
         "dtype": "pl.Int64"
       },
-      "ms2": {
+      "num_ms2": {
         "dtype": "pl.Int64"
       }
     }

{masster-0.3.13.dist-info → masster-0.3.15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: masster
-Version: 0.3.13
+Version: 0.3.15
 Summary: Mass spectrometry data analysis package
 Project-URL: homepage, https://github.com/zamboni-lab/masster
 Project-URL: repository, https://github.com/zamboni-lab/masster
@@ -684,6 +684,7 @@ Requires-Dist: alphabase>=1.0.0
 Requires-Dist: alpharaw>=0.4.8
 Requires-Dist: altair>=5.5.0
 Requires-Dist: bokeh>=3.7.3
+Requires-Dist: cmap>=0.6.2
 Requires-Dist: datashader>=0.18.1
 Requires-Dist: h5py>=3.14.0
 Requires-Dist: holoviews>=1.21.0

{masster-0.3.13.dist-info → masster-0.3.15.dist-info}/RECORD

@@ -3,22 +3,18 @@ masster/_version.py,sha256=ioQa4W_2pWdKSoU7hw7Pn6WMBm3nMuuLKfSR4f8171A,256
 masster/chromatogram.py,sha256=NgPr1uLGJHjRu6PWZZGOrS3pCl7sye1yQCJjlRi9ZSY,19305
 masster/logger.py,sha256=W50V_uh8RSYwGxDrDFhOuj5jpu2tKJyt_16lMw9kQwA,14755
 masster/spectrum.py,sha256=LuDa7qP_JInctzkmxC9c5468opHOholy321KpUgyW2U,47550
-masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.featureXML,sha256=033IjCWBaYVymnPhVHneytilC-XIa6T-6wkeBB0BXvc,10980374
 masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.mzML,sha256=3RS_crLN-aoPSacMYaQ45sxszmp_EcQElrg8tiuAQyA,39741920
-masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.sample5,sha256=T3LxClfe3Uaqx4Rb7lY57e2uWpuJHzPy-VEl3cy4b7k,11109448
 masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.timeseries.data,sha256=01vC6m__Qqm2rLvlTMZoeKIKowFvovBTUnrNl8Uav3E,24576
 masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.wiff,sha256=go5N9gAM1rn4PZAVaoCmdteY9f7YGEM9gyPdSmkQ8PE,1447936
 masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.wiff.scan,sha256=ahi1Y3UhAj9Bj4Q2MlbgPekNdkJvMOoMXVOoR6CeIxc,13881220
 masster/data/examples/2025_01_14_VW_7600_LpMx_DBS_CID_2min_TOP15_030msecMS1_005msecReac_CE35_DBS-ON_3.wiff2,sha256=TFB0HW4Agkig6yht7FtgjUdbXax8jjKaHpSZSvuU5vs,3252224
-masster/docs/SCX_API_Documentation.md,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-masster/docs/SCX_DLL_Analysis.md,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 masster/sample/__init__.py,sha256=HL0m1ept0PMAYUCQtDDnkdOS12IFl6oLAq4TZQz83uY,170
 masster/sample/h5.py,sha256=IdfbdkDgKcij-jMQTxnjW-gsBhb6vwi8w1XXL795yEs,63793
-masster/sample/helpers.py,sha256=U2VyboRdTsQbOefCy7VXh6PlSQtEsR6BK5QF2jGUd94,36208
+masster/sample/helpers.py,sha256=ThmabX_-gXxLFyok-mYMNOr5JErIO_zcrIHwqovXE1c,36541
 masster/sample/lib.py,sha256=l5YdU9TxEWJI0kJxXxrRCxgDDwbzO5zBf1_Qi_HY87w,33556
-masster/sample/load.py,sha256=5Ig94gEWBfdxR86yNwaPj9mdNvPI1nVCp0LXcFJt05U,47577
+masster/sample/load.py,sha256=jeGP-f4nYDJlDt4MoqVEYXJ7ZQEwV8ZtM34S1ATEaNc,47774
 masster/sample/parameters.py,sha256=Gg2KcuNbV_wZ_Wwv93QlM5J19ji0oSIvZLPV1NoBmq0,4456
-masster/sample/plot.py,sha256=C8Y2DLLHw3Bo_U0MPhKMI3AKOkqj0zMUQxasRVWp26U,77964
+masster/sample/plot.py,sha256=mXrQbgBfMhXtfat_xAuliyXNTanW_kezq4hmWQYtFiQ,78445
 masster/sample/processing.py,sha256=-H93MEUysA-B9PB4nU31WFjtaU_flqbu2gY35ce4vVs,57827
 masster/sample/quant.py,sha256=tHNjvUFTdehKR31BXBZnVsBxMD9XJHgaltITOjr71uE,7562
 masster/sample/sample.py,sha256=QxKjXPO5lWRrIq5eDsYzNQPjp0zI_vuPdPlRQe1y3uI,16925
@@ -33,16 +29,16 @@ masster/sample/defaults/get_spectrum_def.py,sha256=o62p31PhGd-LiIkTOzKQhwPtnO2At
 masster/sample/defaults/sample_def.py,sha256=keoXyMyrm_iLgbYqfIbqCpJ3XHBVlNwCNmb5iMQL0iY,14579
 masster/study/__init__.py,sha256=Zspv6U8jFqjkHGYdNdDy1rfUnCSolCzUdgSSg98PRgE,166
 masster/study/export.py,sha256=9Bhz8wpO3ZHdwV0iWSX0E38GS3UfqfAFlW9VN8ht2-Y,28845
-masster/study/h5.py,sha256=UOc4tbeWr8Xa_5Aescz7rMMnkzpu8PSTsOAnTfPv0-E,67109
-masster/study/helpers.py,sha256=qf5_4DVW5X3MI3TgoZes5bNzuIt-r_esSeJoNOO8a6A,92540
+masster/study/h5.py,sha256=Tl_jdV75yOZ5PH76jvMvTdOJdhiup6uINPC04DhcDX0,71815
+masster/study/helpers.py,sha256=PZ0Lb3eFSnEmPQ3sFzHn-cYr_OehHGzTOogfo3r5K-0,133432
 masster/study/helpers_optimized.py,sha256=sd87kNPIEPdMijekXzZWSyeZzJ_DTAW8HQjAry-jVyY,13922
-masster/study/load.py,sha256=bpiyMQU6rtiz4d_EdsI4w1p_T2l8YkRcDmdBUeI3pEk,49571
+masster/study/load.py,sha256=xh-5CX7rCiw_AIY8Wwe_jr4n6BPRSAUIoRr6x6EG8Cs,50192
 masster/study/parameters.py,sha256=0elaF7YspTsB7qyajWAbRNL2VfKlGz5GJLifmO8IGkk,3276
-masster/study/plot.py,sha256=Vvbh19f94X8D4aX5mLVyPso-iC0ZpsyQCe3NTib34VU,68389
-masster/study/processing.py,sha256=EZlzcMmMWcCW0dMdC3tc4r8ii6kncWQC2_Lc_ybfLFc,52286
+masster/study/plot.py,sha256=RZ-ko0ocsXzaPtsa5QzBDX4FIwQkx4lYf5RPLJAH5Ss,76147
+masster/study/processing.py,sha256=ijLaVKZlPJQpcQh_u-Cj6acrSEOaB3vKMNzS9alXdzg,52661
 masster/study/save.py,sha256=YjFEiuiB4OFLVvW_AX4-kgnsbjCWrYZeqF85VNEtbdw,6560
-masster/study/study.py,sha256=ZeCwKgZfAIrvEywxn7HbtBFssWCYxgkRnsNzgmfApyo,31343
-masster/study/study5_schema.json,sha256=Grm2vfi2NnfNfcqKndz3IX9JNyhgwh92T8x-IofLay4,5103
+masster/study/study.py,sha256=IrgkZPFwdgRFyAFrm0KpA7pRXuhvtwzHuVq2ee-3Hj8,31292
+masster/study/study5_schema.json,sha256=k1D6LhHWVioe4GY9-Xg0ro2wp9Z_Oc1MbZ4lBkrAkys,5316
 masster/study/defaults/__init__.py,sha256=m3Z5KXGqsTdh7GjYzZoENERt39yRg0ceVRV1DeCt1P0,610
 masster/study/defaults/align_def.py,sha256=QSJXfe5kAtYp_IN8LUuXjq61IkxT74ml84k5kmmRjqM,19846
 masster/study/defaults/export_def.py,sha256=eXl3h4aoLX88XkHTpqahLd-QZ2gjUqrmjq8IJULXeWo,1203
@@ -53,9 +49,9 @@ masster/study/defaults/find_ms2_def.py,sha256=RL0DFG41wQ05U8UQKUGr3vzSl3mU0m0knQ
 masster/study/defaults/integrate_chrom_def.py,sha256=0MNIWGTjty-Zu-NTQsIweuj3UVqEY3x1x8pK0mPwYak,7264
 masster/study/defaults/integrate_def.py,sha256=Vf4SAzdBfnsSZ3IRaF0qZvWu3gMDPHdgPfMYoPKeWv8,7246
 masster/study/defaults/merge_def.py,sha256=EBsKE3hsAkTEzN9dpdRD5W3_suTKy_WZ_96rwS0uBuE,8572
-masster/study/defaults/study_def.py,sha256=d8mQWIpvWEWI8grPTAcQa4jKTG7QrM98RRgHZVoh134,9519
-masster-0.3.13.dist-info/METADATA,sha256=3UgZrlVxDb1GIwA_8y6-LW0407iuS6s9RNBpOpMmijE,44293
-masster-0.3.13.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-masster-0.3.13.dist-info/entry_points.txt,sha256=ZHguQ_vPmdbpqq2uGtmEOLJfgP-DQ1T0c07Lxh30wc8,58
-masster-0.3.13.dist-info/licenses/LICENSE,sha256=bx5iLIKjgAdYQ7sISn7DsfHRKkoCUm1154sJJKhgqnU,35184
-masster-0.3.13.dist-info/RECORD,,
+masster/study/defaults/study_def.py,sha256=v2V5i5y288gydhMOM78m8u_GaWC2XdjLM5nJP6e17sI,10476
+masster-0.3.15.dist-info/METADATA,sha256=RZMq52mkrHMq9vTYpMRwAcZ-hZkCYkYaZDMm68tcYw0,44320
+masster-0.3.15.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+masster-0.3.15.dist-info/entry_points.txt,sha256=ZHguQ_vPmdbpqq2uGtmEOLJfgP-DQ1T0c07Lxh30wc8,58
+masster-0.3.15.dist-info/licenses/LICENSE,sha256=bx5iLIKjgAdYQ7sISn7DsfHRKkoCUm1154sJJKhgqnU,35184
+masster-0.3.15.dist-info/RECORD,,