Mercurial > hg > lib > markup

annotate python/cluster.py @ 10:73bb35b96624

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
show good hit output
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Thu, 27 May 2021 06:09:01 -0400
parents 3ee329b129c6
children e411408d64ec
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
1 #!/usr/bin/python3
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
2 '''Read a minutor block lockation tsv and resort it to show clusters'''
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
3 import sys
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
5 from util import *
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
6
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
7 if len(sys.argv)==1 or sys.argv[1]=='-h':
7
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
8 print("""Usage: cluster.py [-h] [-c n] [-s] infile.tsv [outfile.tsv]
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
9 n is cluster diameter, default is 5
7
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
10 -s for strict circle distance, reduces number of candidates
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
11 default outfile is [infile]_c[n].tsv""")
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
12 exit(1)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
13 if sys.argv[1]=='-c':
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
14 sys.argv.pop(1)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
15 n=float(sys.argv.pop(1))
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
16 else:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
17 n=5.0
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
18
7
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
19 if sys.argv[1]=='-s':
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
20 sys.argv.pop(1)
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
21 strict=True
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
22 else:
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
23 strict=False
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
24
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
25 infile_name=sys.argv.pop(1)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
26 if len(sys.argv)>1:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
27 outfile_name=sys.argv.pop(1)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
28 else:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
29 outfile_name="%s_c%s.tsv"%(infile_name.split('.')[0],n)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
30
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
31 cc=[]
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
32
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
33 with open(infile_name,'r') as infile:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
34 with open(outfile_name,'w') as outfile:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
35 l=infile.readline().rstrip()
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
36 print(l,file=outfile)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
37 ff=PPAT.split(l)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
38 (nr,ox,oy,oz)=intsMaybe(ff)
5
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
39 home=(float(ox),float(oy),float(oz))
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
40 et=ff[9]
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
41 l=infile.readline().rstrip()
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
42 print(l,file=outfile)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
43 (orad,ymin,ymax)=intsMaybe(PPAT.split(l))
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
44 _=infile.readline()
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
45 for l in infile:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
46 found=False
5
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
47 (_,td,qd)=l.rstrip().split('\t')
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
48 q=[float(i) for i in qd.split(',')]
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
49 td=float(td)
7
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
50 if strict and td>orad:
5
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
51 # Enforce a circular region, not square
7
3ee329b129c6 rename block properties, slightly revise result format
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
52 print(td,orad,file=sys.stderr)
5
672621ab4db4 trim points we check to a circle per radius
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 4
diff changeset
53 break
4
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
54 for c in cc:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
55 for p in c:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
56 if d(p,q)<=n:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
57 c.append(q)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
58 found=True
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
59 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
60 if found:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
61 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
62 if not found:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
63 cc.append([q])
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
64 oc=cc
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
65 cc=[] # lose
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
66 w=0
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
67 ow=-1
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
68 nc=[] # win
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
69 while True:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
70 for i,c in enumerate(oc):
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
71 win=False
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
72 for p in c:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
73 for g in oc[i+1:]:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
74 for q in g:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
75 if d(p,q)<=n:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
76 win=True
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
77 w+=1
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
78 nc.append(c+g)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
79 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
80 if win:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
81 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
82 if win:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
83 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
84 if not win:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
85 cc.append(c)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
86 print(len(cc),len(nc),ow,w,file=sys.stderr)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
87 if ow==w:
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
88 break
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
89 ow=w
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
90 oc=nc
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
91 nc=[]
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
92 for c in sorted(cc,reverse=True,key=lambda x:len(x)):
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
93 print(c,file=outfile)
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
94
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
95
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
96
56508a6033a9 minutor chunk hacking
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
97