/ cutoff determines bootstrap value below which we ignore branches
/ The idea here is that we will take a case where there is
/ a master tree and a lot of individual trees and try to see where
/ there is the most likelihood of interbreeding.
/ For each individual gene, we compare the master with the 
/ individual. If they are incompatible, we try to determine
/ which graph would make the most sense.
/ S1: M1 G1
/ S2: M2 G2
/ S3: M1 G2
/ This seems to suggest M1 --> G
/ and G2 --> M so there is a cycle.
/ We could leave out any of these.
/ Suppose we left out S2.
/ M G
/   M1 G
/     M1 G1
/     M1 G2
/ This would tell us that M G is an ancestor as is M1 G2.
/ M1 G2 is interesting because it's very close to an existent species.
/ Maybe we could register which master nodes are taken.

spit:{[list] ,/ ($list) ,\: (" ")}
spitcomma:{[list] (-1) _ ,/ ($list) ,\: (",")}
spitvert:{[list] (-1) _ ,/ ($list) ,\: ("|")}
spithyphen:{[list] (-1) _ ,/ ($list) ,\: ("-")}

spitstrip:{[list]
  x: stripbrackets'list
  :spit[x]
}

stripbrackets:{[el]
 x: $el
 ii: & ~ x _in\: "[] "
 :` $ x[ii]
}
 
 

processold:{[line]
  if[2 > #line; :()]
  out: ()
  i: 0
  while[i < #line
	char: line[i]
	flag: char  _in ";0123456789"
	if[~flag
		out,: :[char = ","; " "; char]
	]
	i+: 1
  ]
  :out
}

/ cut out any line below bootstrap cutoff but otherwise parse is the
/ same.
/ Also get gene list for yeast genes
processold2:{[line]
  if[2 > #line; :()]
  if[("Y") = line[0]
	namelist,: ,line
	:()
  ]
  out: ()
  doingnumber: 0
  nums: ()
  curnumstring: ()
  i: 0
  while[i < #line
	char: line[i]
	numflag: char  _in "0123456789"
	flag: (char  =  (";")) | numflag
	if[~flag
		out,: :[char = ","; " "; char]
	]
	if[numflag
		if[(~doingnumber); curnumstring: char] / starting number
		if[doingnumber; curnumstring,: char] / continuing number
	]
	if[doingnumber & (~numflag)
		x: 0 $ curnumstring
		nums,: x
	]
	doingnumber: :[numflag; 1; 0]
	i+: 1
  ]
  goodlines,: cutoff < &/nums
  resultlines,: ,out
  :()
}

/ get rid of final paren pair
/ e.g. from  "((((z bkjklj) jk)"
/ to "(((z bkjklj) jk"
huntparens:{[line]
  rline: |line
  i: rline ? ")"
  oline: rline[!i]
  rline: (i+1) _ rline
  i: rline ? "("
  oline,: rline[!i]
  rline: (i+1) _ rline
  oline,: rline
  :|oline
}

/ cut out any line below bootstrap cutoff but otherwise parse is the
/ same.
/ Also get gene list for yeast genes
process:{[line]
  if[2 > #line; :()]
  if[("Y") = line[0]
	namelist,: ,line
	:()
  ]
  out: ()
  doingnumber: 0
  nums: ()
  curnumstring: ()
  origlines,: ,line
  i: 0
  while[i < #line
	char: line[i]
	numflag: char  _in "0123456789"
	flag: (char  =  (";")) | numflag
	if[~flag
		out,: :[char = ","; " "; char]
	]
	if[numflag
		if[(~doingnumber); curnumstring: char] / starting number
		if[doingnumber; curnumstring,: char] / continuing number
	]
	if[doingnumber & (~numflag)
		x: 0 $ curnumstring
		nums,: x
		if[x < cutoff; out: huntparens[out]]
	]
	doingnumber: :[numflag; 1; 0]
	i+: 1
  ]
  / goodlines,: cutoff < &/nums / don't do this
  resultlines,: ,out
  :()
}

add:{[num; line]
  :line, (" [gene"),($num),("]")
}


/ look at the phylogenies and get the ones that are most frequent
findfreq:{[phylogenies; namelist]
  part: = phylogenies
  mynamelist: ()
  firstnamelist: ()
  allnamelist: ()
  k: 0
  while[k < #part
	mynamelist,: ` $ (-1) _ spit[namelist[part[k]]]
	firstnamelist,: ` $ ("["), ((namelist[part[k]])[0]), ("]")
	allnamelist,: ` $ ("["), (spitcomma (namelist[part[k]])), ("]")
	k+: 1
  ]
  firsts: (-1) + *:'part
  uniqs: ?phylogenies
  counts: #:' part
  ii: & 0 < counts / every tree when 0
  "tmpgroups" 1: (stripbrackets'firstnamelist),'(stripbrackets'allnamelist)
  x:(` $ uniqs[ii]),'(counts[ii]),'(firstnamelist[ii]),'(mynamelist[ii])
  ii: > x[;1]
  :x[ii]
}

/ PROCESSING

cutoff: 80 / any bootstrap value less than this, then ignore that split
if[1 > #_i
	` 0: ,"k driver graphlife2 (or some such input file)"
 . "\\\\"
]
a: 0: _i[0]

resultlines: ()
origlines: ()
goodlines: ()
namelist: ()
x: process'a
namelist: (,"master"), namelist
if[0 / we no longer kill lines
  goodlines[0]: 1 / master must be one of them
  ii: & goodlines
  / get the cutoff
  resultlines@: ii
  goodlines@: ii
  namelist@: ii
]
"tmporig" 0: origlines,'namelist
most: findfreq[resultlines; namelist]
most
rest: 1 _ resultlines
 // genenum: !#rest
 // resultlines2: add[;]'[genenum;rest]
 // out: ,"rm result*"
 // i: 0
 // while[i < #resultlines2
	// a: (all; resultlines2[i])
	// (("tmp"),($i)) 0: a
	// out,: ,("k graphlife tmp"),$i
	// i+: 1
 // ]
 // "runall" 0: out

"minigraph" 0: spit'((most[;0]),'(most[;2]))

xx: most[& most[;1] > 1]
yy: spitstrip[xx[;2]]
/ if[~ `"[master]" _in xx[;2]
  / x: ("k graphlife minigraph master "), yy
/ ]
/ if[`"[master]" _in xx[;2]
  x: ("k graphlife minigraph "), yy
/ ]
"minirun" 0: x


\\