diff options
| author | Nicholas Noll <nbnoll@eml.cc> | 2020-07-28 16:39:07 -0700 |
|---|---|---|
| committer | Nicholas Noll <nbnoll@eml.cc> | 2020-07-28 16:39:07 -0700 |
| commit | e1fe2c46651bc53b5ab0006cca07d446b357c79a (patch) | |
| tree | aa5c91c61edb39d67d1bbbb27c992a66b6ad1681 | |
| parent | 8e60f8b5688404fc75d91a6716494782a50b93a8 (diff) | |
fix: must strip gaps upon sequence extraction to have offset correct
| -rw-r--r-- | pangraph/graph.py | 6 | ||||
| -rw-r--r-- | pangraph/sequence.py | 6 | ||||
| -rw-r--r-- | pangraph/tree.py | 1 |
3 files changed, 7 insertions, 6 deletions
diff --git a/pangraph/graph.py b/pangraph/graph.py index 58cb4cc..bc1e9d9 100644 --- a/pangraph/graph.py +++ b/pangraph/graph.py @@ -407,9 +407,9 @@ class Graph(object): # TODO: check that isos is constant along the chain for iso in self.blks[c[0][0]].isolates.keys(): self.seqs[iso].merge(c[0], c[-1], new_blk) - for n in self.seqs[iso].nodes: - if n.blk.id in [e[0] for e in c]: - breakpoint("bad deletion") + # for n in self.seqs[iso].nodes: + # if n.blk.id in [e[0] for e in c]: + # breakpoint("bad deletion") self.blks[new_blk.id] = new_blk for b, _ in c: diff --git a/pangraph/sequence.py b/pangraph/sequence.py index 9374335..b2b5ce7 100644 --- a/pangraph/sequence.py +++ b/pangraph/sequence.py @@ -77,7 +77,7 @@ class Path(object): def sequence(self, verbose=False): seq = "" for n in self.nodes: - s = n.blk.extract(self.name, n.num, strip_gaps=False, verbose=verbose) + s = n.blk.extract(self.name, n.num, strip_gaps=True, verbose=verbose) if n.strand == Strand.Plus: seq += s else: @@ -120,8 +120,6 @@ class Path(object): ids = [n.blk.id for n in self.nodes] try: i, j = ids.index(start[0]), ids.index(stop[0]) - except: - return if self.nodes[i].strand == start[1]: beg, end, s = i, j, Strand.Plus @@ -136,6 +134,8 @@ class Path(object): self.position = np.cumsum([0] + [n.length(self.name) for n in self.nodes]) N += 1 + except: + return def replace(self, blk, tag, new_blks, blk_map): new = [] diff --git a/pangraph/tree.py b/pangraph/tree.py index ba03694..605bac3 100644 --- a/pangraph/tree.py +++ b/pangraph/tree.py @@ -304,6 +304,7 @@ class Tree(object): rec = G.extract(n.name) uncompressed_length += len(orig) if orig != rec: + breakpoint("inconsistency") nerror += 1 with open("test.fa", "w+") as out: |