another hotfix for a rare aligned insertions bug

Gaius-Augustus · Jun 14, 2023 · 39c5c1d · 39c5c1d
1 parent a72c6b9
commit 39c5c1d
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 17 deletions.
diff --git a/learnMSA/_version.py b/learnMSA/_version.py
@@ -1 +1 @@
-__version__ = "1.2.3"
+__version__ = "1.2.4"
diff --git a/learnMSA/msa_hmm/AlignInsertions.py b/learnMSA/msa_hmm/AlignInsertions.py
@@ -16,9 +16,7 @@ def find_long_insertions_and_write_slice(fasta_file, lens, starts, name, directo
         """
         at_least_t = lens >= t
         lengths = lens[at_least_t]
-        if lengths.size > k:
-            if verbose:
-                print(f"Long insertions found at {name}: {lengths.size}.")
+        if lengths.size > 1:
             which = np.squeeze(np.argwhere(at_least_t))
             start = starts[at_least_t]
             filename = f"{directory}/slice_{name}"
@@ -38,7 +36,10 @@ def find_long_insertions_and_write_slice(fasta_file, lens, starts, name, directo
                         slice_file.write(">"+fasta_file.seq_ids[which[j]]+"\n")
                         slice_file.write(segment+"\n")
             which = np.delete(which, to_delete)
-            return (which, filename)
+            if which.size > k:
+                if verbose:
+                    print(f"Long insertions found at {name}: {which.size}.")
+                return (which, filename)
 
 
 def make_aligned_insertions(am, directory, method="famsa", threads=0, verbose=True):

diff --git a/test/data/egf.out.fasta b/test/data/egf.out.fasta
@@ -1,24 +1,24 @@
 >1ixa
-..vdgdqCESN...PCLNGGSC..-..KD...DIN..S.YECWCPFGFEGKNcel....................
+..vdgdqCESN...PCLNGGSCK.--.DDIN..S.YECWCPFGFEGKNcel....................
 >1apo
-..kdgdqCEGH...PCLNQGHC..-..KD...GIG..D.YTCTCAEGFEGKNcefstr.................
+..kdgdqCEGH...PCLNQGHCK.--.DGIG..D.YTCTCAEGFEGKNcefstr.................
 >1urk
-......qVPSNc..DCLNGGTC..-..VSnkyFSN..I.HWCNCPKKFGGQHceidk..................
+.......QVPSnc.DCLNGGTCVsNK.-YFS..NiHWCNCPKKFGGQHceidk..................
 >1fsb
-....tasCQDM...SCSKQGEC..-..LE...TIG..N.YTCSCYPGFYGPEceyvre.................
+....tasCQDM...SCSKQGECL.--.ETIG..N.YTCSCYPGFYGPEceyvre.................
 >1esl
-......aCTNT...SCSGHGEC..-..VE...TIN..N.YTCKCDPGFSGLKceqiv..................
+......aCTNT...SCSGHGECV.--.ETIN..N.YTCKCDPGFSGLKceqiv..................
 >1hre
-gtshlvkCAEKektFCVNGGEC..FmvKD...LSNpsR.YLCKCQPGFTGARctenvpmkvqnqekaeelyqk..
+gtshlvkCAEKektFCVNGGECF.MVkDLSNpsR.YLCKCQPGFTGARctenvpmkvqnqekaeelyqk..
 >1epi
-..nsypgCPSSydgYCLNGGVC..Mh.IE...SLD..S.YTCNCVIGYSGDRcqtrdlrwwelr...........
+..nsypgCPSSydgYCLNGGVCM.HI.ESLD..S.YTCNCVIGYSGDRcqtrdlrwwelr...........
 >4tgf
-vvshfndCPDShtqFCFH-GTCrfL..-V...QED..K.PACVCHSGYVGARcehadlla...............
+vvshfndCPDShtqFCFH-GTCRfLV.-QED..K.PACVCHSGYVGARcehadlla...............
 >1hcgb
-.......CSLD...NGDCDQFC..-..HE...EQN..S.VVCSCARGYTLADngkaciptgpypcgkqtler...
+.......CSLD...NGDCDQFCH.--.EEQN..S.VVCSCARGYTLADngkaciptgpypcgkqtler...
 >1dan1
-....gdqCASS...PCQNGGSC..-..KD...QLQ..S.YICFCLPAFEGRNcethkd.................
+....gdqCASS...PCQNGGSCK.--.DQLQ..S.YICFCLPAFEGRNcethkd.................
 >1dan2
-...dqliCVNE...NGGCEQYC..-..SD...HTG..TkRSCRCHEGYSLLAdgvsctptveypcgkipile...
+...dqliCVNE...NGGCEQYCS.--.DHTG..TkRSCRCHEGYSLLAdgvsctptveypcgkipile...
 >1rfnb
-.....mtCNIK...NGRCEQFC..-..KN...SAD..NkVVCSCTEGYRLAEnqkscepavpfpcgrvsvsqtsk
+.....mtCNIK...NGRCEQFCK.NS.-ADN..K.VVCSCTEGYRLAEnqkscepavpfpcgrvsvsqtsk