ENH: add annotations-to-metadata transformer (#318)

qiime2 · Apr 4, 2024 · d24a69d · d24a69d
1 parent 823b5a4
commit d24a69d
Show file tree

Hide file tree

Showing 8 changed files with 239 additions and 12 deletions.
diff --git a/q2_types/feature_data_mag/_format.py b/q2_types/feature_data_mag/_format.py
@@ -49,14 +49,14 @@ def feature_dict(self, relative=False):
             if not pattern.match(path.name):
                 continue
 
-            id = path.stem
+            _id = path.stem
             absolute_path = path.absolute()
             if relative:
-                ids[id] = str(
+                ids[_id] = str(
                     absolute_path.relative_to(self.path.absolute())
                 )
             else:
-                ids[id] = str(absolute_path)
+                ids[_id] = str(absolute_path)
 
         return dict(sorted(ids.items()))
 
@@ -65,14 +65,27 @@ def feature_dict(self, relative=False):
 
 
 class OrthologAnnotationDirFmt(model.DirectoryFormat):
-    annotations = model.FileCollection(
-        r'.+\.annotations',
-        format=OrthologFileFmt
-    )
+    pathspec = r'.+\.annotations'
+    annotations = model.FileCollection(pathspec, format=OrthologFileFmt)
 
     @annotations.set_path_maker
     def annotations_path_maker(self, file_name):
         return file_name.split(sep="_")[0]
 
+    def annotation_dict(self, relative=False) -> dict:
+        ids = {}
+        for path in self.path.iterdir():
+            if re.compile(self.pathspec).match(path.name):
+                _id = re.sub('.emapper$', '', path.stem)
+                absolute_path = path.absolute()
+                if relative:
+                    ids[_id] = str(
+                        absolute_path.relative_to(self.path.absolute())
+                    )
+                else:
+                    ids[_id] = str(absolute_path)
+
+        return dict(sorted(ids.items()))
+
 
 plugin.register_formats(OrthologAnnotationDirFmt)
diff --git a/q2_types/feature_data_mag/_transformer.py b/q2_types/feature_data_mag/_transformer.py
@@ -5,17 +5,22 @@
 #
 # The full license is in the file LICENSE, distributed with this software.
 # ----------------------------------------------------------------------------
+from io import StringIO
+
+import uuid
 
 import collections.abc
 import glob
 import os.path
+
+import qiime2
 from itertools import repeat
 
 import pandas as pd
 import skbio
 from q2_types.feature_data._transformer import _fastaformats_to_series
 
-from . import MAGSequencesDirFmt
+from . import MAGSequencesDirFmt, OrthologAnnotationDirFmt
 from ..plugin_setup import plugin
 
 CONSTRUCTORS = {
@@ -89,3 +94,84 @@ def _5(data: MAGIterator) -> MAGSequencesDirFmt:
         with open(fp, 'a') as fin:
             skbio.io.write(seq, format='fasta', into=fin)
     return result
+
+
+def _is_valid_uuid4(uuid_string: str):
+    """
+    Check if a given string is a valid UUID version 4.
+
+    This function checks if the provided string is a valid UUID version 4.
+    The only purpose of doing that here is to identify whether provided
+    string was a MAG ID (UUID4) or a sample ID. For that reason, we don't
+    print any statements or raise any exceptions.
+
+    Parameters:
+    uuid_string (str): The string to check for UUID version 4 validity.
+
+    Returns:
+    bool: True if the string is a valid UUID version 4, False otherwise.
+    """
+    try:
+        uuid_obj = uuid.UUID(uuid_string, version=4)
+    except ValueError:
+        return False
+
+    return str(uuid_obj) == uuid_string
+
+
+def _reshuffle_columns(df: pd.DataFrame):
+    if 'MAG' in df.columns:
+        col = 'MAG'
+    elif 'Sample' in df.columns:
+        col = 'Sample'
+    else:
+        return df
+
+    cols = df.columns.tolist()
+    cols.remove(col)
+    cols.insert(0, col)
+    df = df[cols]
+    return df
+
+
+def _annotations_to_dataframe(
+        data: OrthologAnnotationDirFmt
+) -> pd.DataFrame:
+    annotations = data.annotation_dict()
+    dfs = []
+    for _id, path in annotations.items():
+        # we need to ignore the ## comments at
+        # the beginning and end of the file
+        with open(path, 'r') as f:
+            lines = [line for line in f if not line.startswith('##')]
+
+        df = pd.read_csv(StringIO('\n'.join(lines)), sep='\t', index_col=0)
+        if _is_valid_uuid4(_id):
+            df['MAG'] = _id
+        else:
+            df['Sample'] = _id
+
+        dfs.append(df)
+
+    df = pd.concat(dfs)
+
+    # to satisfy QIIME2's particular requirements
+    df.reset_index(drop=False, inplace=True)
+    df.index = df.index.astype(str)
+    df.index.rename('id', inplace=True)
+
+    # reshuffle columns for nicer display
+    df = _reshuffle_columns(df)
+
+    return df
+
+
+@plugin.register_transformer
+def _7(data: OrthologAnnotationDirFmt) -> pd.DataFrame:
+    return _annotations_to_dataframe(data)
+
+
+@plugin.register_transformer
+def _8(data: OrthologAnnotationDirFmt) -> qiime2.Metadata:
+    annotations = _annotations_to_dataframe(data)
+    return qiime2.Metadata(annotations)
diff --git a/...ts/data/ortholog_annotation_mags/d33dbcbe-eedd-4a19-8b93-e76a15ce107f.emapper.annotations b/...ts/data/ortholog_annotation_mags/d33dbcbe-eedd-4a19-8b93-e76a15ce107f.emapper.annotations
@@ -0,0 +1,12 @@
+## Tue Apr  2 21:44:33 2024
+## emapper-2.1.12-31d78ce
+## /home/someone/miniconda3/envs/q2-shotgun/bin/emapper.py -m no_search --annotate_hits_table /scratch/someone/tmp/qiime2/someone/data/1678d775-94b8-4b87-8c40-937d80e23e65/data/3af39f47-e90a-46f2-9b5f-b236ae6551f0.emapper.seed_orthologs --data_dir /scratch/someone/tmp/qiime2/someone/data/499c1a8d-66be-4e80-816b-d142608fa9a6/data -o 3af39f47-e90a-46f2-9b5f-b236ae6551f0 --output_dir /scratch/someone/tmp/q2-OrthologAnnotationDirFmt-lzc_e07g --cpu 11
+##
+#query	seed_ortholog	evalue	score	eggNOG_OGs	max_annot_lvl	COG_category	Description	Preferred_name	GOs	EC	KEGG_ko	KEGG_Pathway	KEGG_Module	KEGG_Reaction	KEGG_rclass	BRITE	KEGG_TC	CAZy	BiGG_Reaction	PFAMs
+k141_710_0	83332.Rv1754c	4.94e-270	827.0	COG4409@1|root,COG4409@2|Bacteria,2GKYI@201174|Actinobacteria,233HH@1762|Mycobacteriaceae	201174|Actinobacteria	G	Domain of unknown function (DUF4185)	-	-	-	-	-	-	-	-	-	-	-	-	DUF4185
+k141_710_1	216594.MMAR_4577	3.36e-233	727.0	COG0515@1|root,COG3391@1|root,COG0515@2|Bacteria,COG3391@2|Bacteria,2GNXG@201174|Actinobacteria,23296@1762|Mycobacteriaceae	201174|Actinobacteria	KLT	serine threonine protein kinase	pknD	GO:0003674,GO:0003824,GO:0004672,GO:0004674,GO:0005488,GO:0005575,GO:0005576,GO:0005618,GO:0005622,GO:0005623,GO:0005737,GO:0005829,GO:0005886,GO:0005887,GO:0006464,GO:0006468,GO:0006793,GO:0006796,GO:0006807,GO:0006950,GO:0007154,GO:0008150,GO:0008152,GO:0009267,GO:0009405,GO:0009605,GO:0009889,GO:0009890,GO:0009892,GO:0009987,GO:0009991,GO:0010565,GO:0016020,GO:0016021,GO:0016036,GO:0016301,GO:0016310,GO:0016740,GO:0016772,GO:0016773,GO:0019216,GO:0019217,GO:0019222,GO:0019538,GO:0030145,GO:0030312,GO:0031224,GO:0031226,GO:0031323,GO:0031324,GO:0031326,GO:0031327,GO:0031667,GO:0031668,GO:0031669,GO:0032091,GO:0033554,GO:0036211,GO:0042304,GO:0042594,GO:0043085,GO:0043086,GO:0043167,GO:0043169,GO:0043170,GO:0043393,GO:0043412,GO:0044092,GO:0044093,GO:0044237,GO:0044238,GO:0044260,GO:0044267,GO:0044419,GO:0044424,GO:0044425,GO:0044444,GO:0044459,GO:0044464,GO:0045717,GO:0045833,GO:0045922,GO:0046872,GO:0046890,GO:0046914,GO:0048519,GO:0048523,GO:0050789,GO:0050790,GO:0050794,GO:0050896,GO:0051055,GO:0051098,GO:0051100,GO:0051704,GO:0051716,GO:0062012,GO:0062014,GO:0065007,GO:0065009,GO:0071496,GO:0071704,GO:0071944,GO:0080090,GO:0140096,GO:1901564	2.7.11.1	ko:K08884,ko:K12132	-	-	-	-	ko00000,ko01000,ko01001	-	-	-	NHL,Pkinase,PknH_C
+k141_710_2	525368.HMPREF0591_4743	9.84e-212	650.0	COG3804@1|root,COG3804@2|Bacteria,2GKE3@201174|Actinobacteria,232FQ@1762|Mycobacteriaceae	201174|Actinobacteria	S	dihydrodipicolinate reductase	-	-	1.4.1.12,1.4.1.26	ko:K21672	ko00310,ko00330,ko00472,map00310,map00330,map00472	-	R02825,R04200,R04201,R04687,R04688	RC00249,RC00790	ko00000,ko00001,ko01000	-	-	-	DapB_N
+k141_710_3	557599.MKAN_09155	1.65e-200	619.0	COG0226@1|root,COG0226@2|Bacteria,2GJXD@201174|Actinobacteria,233MY@1762|Mycobacteriaceae	201174|Actinobacteria	P	Part of the ABC transporter complex PstSACB involved in phosphate import	pstS	GO:0002682,GO:0002683,GO:0005575,GO:0005576,GO:0005618,GO:0005623,GO:0005886,GO:0006950,GO:0007154,GO:0008150,GO:0009267,GO:0009605,GO:0009607,GO:0009987,GO:0009991,GO:0016020,GO:0016036,GO:0030312,GO:0031347,GO:0031348,GO:0031667,GO:0031668,GO:0031669,GO:0033554,GO:0035821,GO:0040007,GO:0042594,GO:0043207,GO:0044003,GO:0044110,GO:0044116,GO:0044117,GO:0044403,GO:0044413,GO:0044414,GO:0044419,GO:0044464,GO:0045088,GO:0045824,GO:0048519,GO:0048583,GO:0048585,GO:0050776,GO:0050777,GO:0050789,GO:0050896,GO:0051701,GO:0051704,GO:0051707,GO:0051716,GO:0051817,GO:0051832,GO:0051833,GO:0052031,GO:0052037,GO:0052167,GO:0052170,GO:0052173,GO:0052200,GO:0052255,GO:0052261,GO:0052306,GO:0052309,GO:0052552,GO:0052553,GO:0052561,GO:0052562,GO:0052564,GO:0052572,GO:0065007,GO:0071496,GO:0071944,GO:0075136,GO:0080134	-	ko:K02040	ko02010,ko02020,ko05152,map02010,map02020,map05152	M00222	-	-	ko00000,ko00001,ko00002,ko02000	3.A.1.7	-	-	PBP_like_2
+## 1572 queries scanned
+## Total time (seconds): 0.3267531394958496
+## Rate: 4810.97 q/s
diff --git a/...ts/data/ortholog_annotation_mags/e6408340-b308-4325-83a1-ea406f3e19ad.emapper.annotations b/...ts/data/ortholog_annotation_mags/e6408340-b308-4325-83a1-ea406f3e19ad.emapper.annotations
@@ -0,0 +1,15 @@
+## Tue Apr  2 21:44:39 2024
+## emapper-2.1.12-31d78ce
+## /home/someone/miniconda3/envs/q2-shotgun/bin/emapper.py -m no_search --annotate_hits_table /scratch/someone/tmp/qiime2/someone/data/aa200c09-5719-4dca-a545-15e8b39d8937/data/b46723b9-9768-4db3-aed7-1b4978987aef.emapper.seed_orthologs --data_dir /scratch/someone/tmp/qiime2/someone/data/499c1a8d-66be-4e80-816b-d142608fa9a6/data -o b46723b9-9768-4db3-aed7-1b4978987aef --output_dir /scratch/someone/tmp/q2-OrthologAnnotationDirFmt-uwc_7tfv --cpu 11
+##
+#query	seed_ortholog	evalue	score	eggNOG_OGs	max_annot_lvl	COG_category	Description	Preferred_name	GOs	EC	KEGG_ko	KEGG_Pathway	KEGG_Module	KEGG_Reaction	KEGG_rclass	BRITE	KEGG_TC	CAZy	BiGG_Reaction	PFAMs
+k141_6_0	216594.MMAR_3837	1.03e-31	126.0	2A1QU@1|root,30PZC@2|Bacteria,2I8J2@201174|Actinobacteria,233GM@1762|Mycobacteriaceae	201174|Actinobacteria	M	Catalyzes the hydrolysis of cutin, a polyester that forms the structure of plant cuticle	-	-	3.1.1.74	ko:K08095	-	-	-	-	ko00000,ko01000	-	-	-	Cutinase
+k141_14_0	272631.ML1658	1.03e-169	514.0	COG0266@1|root,COG0266@2|Bacteria,2GJNT@201174|Actinobacteria,23633@1762|Mycobacteriaceae	201174|Actinobacteria	L	Involved in base excision repair of DNA damaged by oxidation or by mutagenic agents. Acts as DNA glycosylase that recognizes and removes damaged bases. Has a preference for oxidized purines, such as 7,8-dihydro-8-oxoguanine (8-oxoG). Has AP (apurinic apyrimidinic) lyase activity and introduces nicks in the DNA strand. Cleaves the DNA backbone by beta-delta elimination to generate a single-strand break at the site of the removed base with both 3'- and 5'-phosphates	fpg	GO:0000702,GO:0003674,GO:0003676,GO:0003677,GO:0003690,GO:0003824,GO:0005488,GO:0005575,GO:0005623,GO:0005886,GO:0006139,GO:0006259,GO:0006281,GO:0006284,GO:0006725,GO:0006807,GO:0006950,GO:0006974,GO:0006979,GO:0008150,GO:0008152,GO:0008534,GO:0009987,GO:0016020,GO:0016787,GO:0016798,GO:0016799,GO:0019104,GO:0033554,GO:0034039,GO:0034641,GO:0043170,GO:0044237,GO:0044238,GO:0044260,GO:0044464,GO:0046483,GO:0050896,GO:0051716,GO:0071704,GO:0071944,GO:0090304,GO:0097159,GO:0140097,GO:1901360,GO:1901363	3.2.2.23,4.2.99.18	ko:K10563	ko03410,map03410	-	-	-	ko00000,ko00001,ko01000,ko03400	-	-	-	Fapy_DNA_glyco,H2TH,zf-FPG_IleRS
+k141_14_1	272631.ML1659	5.03e-147	452.0	COG0571@1|root,COG0571@2|Bacteria,2GKER@201174|Actinobacteria,232ZI@1762|Mycobacteriaceae	201174|Actinobacteria	J	Digests double-stranded RNA. Involved in the processing of primary rRNA transcript to yield the immediate precursors to the large and small rRNAs (23S and 16S). Processes some mRNAs, and tRNAs when they are encoded in the rRNA operon. Processes pre- crRNA and tracrRNA of type II CRISPR loci if present in the organism	rnc	GO:0003674,GO:0003676,GO:0003723,GO:0003725,GO:0003824,GO:0004518,GO:0004519,GO:0004521,GO:0004525,GO:0004540,GO:0005488,GO:0006139,GO:0006396,GO:0006725,GO:0006807,GO:0008150,GO:0008152,GO:0009987,GO:0010467,GO:0016070,GO:0016787,GO:0016788,GO:0016891,GO:0016893,GO:0032296,GO:0034641,GO:0040007,GO:0043170,GO:0044237,GO:0044238,GO:0046483,GO:0071704,GO:0090304,GO:0090305,GO:0090501,GO:0090502,GO:0097159,GO:0140098,GO:1901360,GO:1901363	3.1.26.3	ko:K03685	ko03008,ko05205,map03008,map05205	-	-	-	ko00000,ko00001,ko01000,ko03009,ko03019,ko03036	-	-	-	Ribonucleas_3_3,dsrm
+k141_14_2	272631.ML1660	1.01e-106	340.0	COG1399@1|root,COG1399@2|Bacteria,2GJTS@201174|Actinobacteria,232I7@1762|Mycobacteriaceae	201174|Actinobacteria	S	Uncharacterized ACR, COG1399	-	GO:0008150,GO:0040007	-	ko:K07040	-	-	-	-	ko00000	-	-	-	DUF177
+k141_14_3	557599.MKAN_24055	4.04e-34	134.0	COG3599@1|root,COG3599@2|Bacteria,2I2NA@201174|Actinobacteria,234IY@1762|Mycobacteriaceae	201174|Actinobacteria	D	Cell division initiation protein	-	GO:0008150,GO:0040007	-	-	-	-	-	-	-	-	-	-	-
+k141_246_0	272631.ML1651	0.0	1826.0	COG2609@1|root,COG2609@2|Bacteria,2GJRE@201174|Actinobacteria,2353T@1762|Mycobacteriaceae	201174|Actinobacteria	C	Component of the pyruvate dehydrogenase (PDH) complex, that catalyzes the overall conversion of pyruvate to acetyl-CoA and CO(2)	aceE	GO:0000287,GO:0003674,GO:0003824,GO:0004738,GO:0004739,GO:0005488,GO:0005575,GO:0005576,GO:0005618,GO:0005622,GO:0005623,GO:0005737,GO:0005886,GO:0008150,GO:0008152,GO:0016020,GO:0016491,GO:0016624,GO:0016903,GO:0030312,GO:0032991,GO:0043167,GO:0043169,GO:0044424,GO:0044444,GO:0044464,GO:0045254,GO:0046872,GO:0055114,GO:0071944,GO:1902494,GO:1990204	1.2.4.1	ko:K00163	ko00010,ko00020,ko00620,ko01100,ko01110,ko01120,ko01130,ko01200,map00010,map00020,map00620,map01100,map01110,map01120,map01130,map01200	M00307	R00014,R00209,R01699,R03270	RC00004,RC00027,RC00627,RC02742,RC02744,RC02882	br01601,ko00000,ko00001,ko00002,ko01000	-	-	-	Transketolase_N
+k141_246_1	272631.ML1630	0.0	1781.0	COG1391@1|root,COG1391@2|Bacteria,2GJ91@201174|Actinobacteria,233Q9@1762|Mycobacteriaceae	201174|Actinobacteria	H	Involved in the regulation of glutamine synthetase GlnA, a key enzyme in the process to assimilate ammonia. When cellular nitrogen levels are high, the C-terminal adenylyl transferase (AT) inactivates GlnA by covalent transfer of an adenylyl group from ATP to specific tyrosine residue of GlnA, thus reducing its activity. Conversely, when nitrogen levels are low, the N-terminal adenylyl removase (AR) activates GlnA by removing the adenylyl group by phosphorolysis, increasing its activity. The regulatory region of GlnE binds the signal transduction protein PII (GlnB) which indicates the nitrogen status of the cell	glnE	GO:0000820,GO:0003674,GO:0003824,GO:0005575,GO:0005618,GO:0005622,GO:0005623,GO:0005737,GO:0005829,GO:0005886,GO:0006521,GO:0008150,GO:0008882,GO:0010565,GO:0016020,GO:0016740,GO:0016772,GO:0016779,GO:0019222,GO:0030312,GO:0031323,GO:0033238,GO:0040007,GO:0042221,GO:0044424,GO:0044444,GO:0044464,GO:0050789,GO:0050794,GO:0050896,GO:0051171,GO:0060359,GO:0062012,GO:0065007,GO:0070566,GO:0071944,GO:0080090,GO:1901698	2.7.7.42,2.7.7.89	ko:K00982	-	-	-	-	ko00000,ko01000	-	-	-	GlnD_UR_UTase,GlnE
+## 1572 queries scanned
+## Total time (seconds): 0.3267531394958496
+## Rate: 4810.97 q/s
diff --git a/.../feature_data_mag/tests/data/ortholog_annotation_samples/test_output1.emapper.annotations b/.../feature_data_mag/tests/data/ortholog_annotation_samples/test_output1.emapper.annotations
@@ -0,0 +1,12 @@
+## Tue Apr  2 21:44:33 2024
+## emapper-2.1.12-31d78ce
+## /home/someone/miniconda3/envs/q2-shotgun/bin/emapper.py -m no_search --annotate_hits_table /scratch/someone/tmp/qiime2/someone/data/1678d775-94b8-4b87-8c40-937d80e23e65/data/3af39f47-e90a-46f2-9b5f-b236ae6551f0.emapper.seed_orthologs --data_dir /scratch/someone/tmp/qiime2/someone/data/499c1a8d-66be-4e80-816b-d142608fa9a6/data -o 3af39f47-e90a-46f2-9b5f-b236ae6551f0 --output_dir /scratch/someone/tmp/q2-OrthologAnnotationDirFmt-lzc_e07g --cpu 11
+##
+#query	seed_ortholog	evalue	score	eggNOG_OGs	max_annot_lvl	COG_category	Description	Preferred_name	GOs	EC	KEGG_ko	KEGG_Pathway	KEGG_Module	KEGG_Reaction	KEGG_rclass	BRITE	KEGG_TC	CAZy	BiGG_Reaction	PFAMs
+k141_710_0	83332.Rv1754c	4.94e-270	827.0	COG4409@1|root,COG4409@2|Bacteria,2GKYI@201174|Actinobacteria,233HH@1762|Mycobacteriaceae	201174|Actinobacteria	G	Domain of unknown function (DUF4185)	-	-	-	-	-	-	-	-	-	-	-	-	DUF4185
+k141_710_1	216594.MMAR_4577	3.36e-233	727.0	COG0515@1|root,COG3391@1|root,COG0515@2|Bacteria,COG3391@2|Bacteria,2GNXG@201174|Actinobacteria,23296@1762|Mycobacteriaceae	201174|Actinobacteria	KLT	serine threonine protein kinase	pknD	GO:0003674,GO:0003824,GO:0004672,GO:0004674,GO:0005488,GO:0005575,GO:0005576,GO:0005618,GO:0005622,GO:0005623,GO:0005737,GO:0005829,GO:0005886,GO:0005887,GO:0006464,GO:0006468,GO:0006793,GO:0006796,GO:0006807,GO:0006950,GO:0007154,GO:0008150,GO:0008152,GO:0009267,GO:0009405,GO:0009605,GO:0009889,GO:0009890,GO:0009892,GO:0009987,GO:0009991,GO:0010565,GO:0016020,GO:0016021,GO:0016036,GO:0016301,GO:0016310,GO:0016740,GO:0016772,GO:0016773,GO:0019216,GO:0019217,GO:0019222,GO:0019538,GO:0030145,GO:0030312,GO:0031224,GO:0031226,GO:0031323,GO:0031324,GO:0031326,GO:0031327,GO:0031667,GO:0031668,GO:0031669,GO:0032091,GO:0033554,GO:0036211,GO:0042304,GO:0042594,GO:0043085,GO:0043086,GO:0043167,GO:0043169,GO:0043170,GO:0043393,GO:0043412,GO:0044092,GO:0044093,GO:0044237,GO:0044238,GO:0044260,GO:0044267,GO:0044419,GO:0044424,GO:0044425,GO:0044444,GO:0044459,GO:0044464,GO:0045717,GO:0045833,GO:0045922,GO:0046872,GO:0046890,GO:0046914,GO:0048519,GO:0048523,GO:0050789,GO:0050790,GO:0050794,GO:0050896,GO:0051055,GO:0051098,GO:0051100,GO:0051704,GO:0051716,GO:0062012,GO:0062014,GO:0065007,GO:0065009,GO:0071496,GO:0071704,GO:0071944,GO:0080090,GO:0140096,GO:1901564	2.7.11.1	ko:K08884,ko:K12132	-	-	-	-	ko00000,ko01000,ko01001	-	-	-	NHL,Pkinase,PknH_C
+k141_710_2	525368.HMPREF0591_4743	9.84e-212	650.0	COG3804@1|root,COG3804@2|Bacteria,2GKE3@201174|Actinobacteria,232FQ@1762|Mycobacteriaceae	201174|Actinobacteria	S	dihydrodipicolinate reductase	-	-	1.4.1.12,1.4.1.26	ko:K21672	ko00310,ko00330,ko00472,map00310,map00330,map00472	-	R02825,R04200,R04201,R04687,R04688	RC00249,RC00790	ko00000,ko00001,ko01000	-	-	-	DapB_N
+k141_710_3	557599.MKAN_09155	1.65e-200	619.0	COG0226@1|root,COG0226@2|Bacteria,2GJXD@201174|Actinobacteria,233MY@1762|Mycobacteriaceae	201174|Actinobacteria	P	Part of the ABC transporter complex PstSACB involved in phosphate import	pstS	GO:0002682,GO:0002683,GO:0005575,GO:0005576,GO:0005618,GO:0005623,GO:0005886,GO:0006950,GO:0007154,GO:0008150,GO:0009267,GO:0009605,GO:0009607,GO:0009987,GO:0009991,GO:0016020,GO:0016036,GO:0030312,GO:0031347,GO:0031348,GO:0031667,GO:0031668,GO:0031669,GO:0033554,GO:0035821,GO:0040007,GO:0042594,GO:0043207,GO:0044003,GO:0044110,GO:0044116,GO:0044117,GO:0044403,GO:0044413,GO:0044414,GO:0044419,GO:0044464,GO:0045088,GO:0045824,GO:0048519,GO:0048583,GO:0048585,GO:0050776,GO:0050777,GO:0050789,GO:0050896,GO:0051701,GO:0051704,GO:0051707,GO:0051716,GO:0051817,GO:0051832,GO:0051833,GO:0052031,GO:0052037,GO:0052167,GO:0052170,GO:0052173,GO:0052200,GO:0052255,GO:0052261,GO:0052306,GO:0052309,GO:0052552,GO:0052553,GO:0052561,GO:0052562,GO:0052564,GO:0052572,GO:0065007,GO:0071496,GO:0071944,GO:0075136,GO:0080134	-	ko:K02040	ko02010,ko02020,ko05152,map02010,map02020,map05152	M00222	-	-	ko00000,ko00001,ko00002,ko02000	3.A.1.7	-	-	PBP_like_2
+## 1572 queries scanned
+## Total time (seconds): 0.3267531394958496
+## Rate: 4810.97 q/s