!*************************************************************************************************************
! Parse-Blast: Parses output from Blast implemented by TIGR CMR. 
!
! Saves all hits of a query (one sequence) on a database (Subjcts)
!
!           Outputs (comma delimited):
!              QueryName   Name of the query 
!              QueryLen    Length of Query
!	       QueryFrom   Position in query where the hit starts
!	       QueryTo     Position in query where the hit ends
!              SubjctName  Name of the subject
!              SubjctLen   Length of hit sequence
!	       Orientation Orientation of the hit (c if it's plus/minus, d if it's plus/plus)
!	       SubjectFrom Position in subject where the hit starts
!	       SubjectTo   Position in subject where the hit ends
!              EValue1     E value of hit
!	       % identity  Percent identity between query and subject of the hit
!
!           If hit not reported, then all values = 0
!*************************************************************************************************************

!*************************************************************************************************************
!
!	Libraries and Declarations
!
!*************************************************************************************************************

  LIBRARY "utilities.trc"              ! Contains Explode
  DECLARE FUNCTION DelChar$

!*************************************************************************************************************
!
!	File Definitions
!
!*************************************************************************************************************

  LET blastFile$ = "alr2168vs7120CMR.txt"
  LET blastFile = 1
  OPEN #BlastFile: NAME BlastFile$, ORG text, CREATE old, ACCESS input

  LET output = 2
  LET outfile$ = "test.txt"
  OPEN #output: NAME outfile$, ORG text, CREATE NewOld, ACCESS OutIn
  ERASE #output
  SET #output: MARGIN Maxnum         ! Allows printing lines of any length

!*************************************************************************************************************
!
!	Initializes global variables 
!
!*************************************************************************************************************

  LET true = 1
  LET false = 0
  LET comma$ = ","
  DIM field$(1)               ! Used by Explode

!*************************************************************************************************************
!
!	Main program
!
!*************************************************************************************************************

   PRINT "Beginning to read file"
   CALL Find_next_query 

   DO WHILE query_found = true
     CALL Evaluate_query 
     CALL Find_next_target
      
     DO WHILE target_found = true
       CALL Evaluate_target
       CALL Find_next_hit

       DO WHILE hit_found = true
         CALL Evaluate_score
         CALL Find_seq
       LOOP 

       IF target_found = false AND query_found = false THEN CALL Find_next_target

     LOOP 

     IF query_found = false THEN CALL Find_next_query

  LOOP 

!*************************************************************************************************************
!
!	SUBROUTINES and FUNCTIONS
!
!*************************************************************************************************************

SUB InputBlast
  LINE INPUT #BlastFile: line$
!  print line$
END SUB

!*************************************************************************************************************
!
!	Deals with the query section of the blast output file.  Get the name and length of the query out
!
!*************************************************************************************************************

SUB Find_next_query
  LET query_found = false
  DO WHILE MORE #BlastFile
    CALL InputBlast
    IF line$[1:6] = "Query=" THEN 
      LET query_found = true
      EXIT DO
    END IF
  LOOP
END SUB

SUB Evaluate_query
  CALL Explode(line$,field$,"= ")
  IF Size(field$) < 1 THEN 
    CAUSE ERROR 11, "Bad line in Evaluate_query"
  ELSE IF Size(field$) < 2 THEN
    LET query_name$ = "(unnamed)"
  ELSE
    LET query_name$ = field$(2) 
  END IF
  CALL InputBlast
  CALL Explode(line$,field$," ()")
  LET query_length$ = field$(1)
  PRINT "QUERY: "; query_name$
END SUB

!*************************************************************************************************************
!
!	Deals with the database section of the blast output file.  get the name and length of the subject out
!
!*************************************************************************************************************

SUB Find_next_target
  LET target_found = false

  DO WHILE MORE #BlastFile
    CALL InputBlast
    IF line$[1:1] = ">" THEN 
      LET target_found = true
      EXIT DO
    ELSE IF line$[1:6] = " *****" THEN 
      LET target_found = false
      LET query_found = false
      EXIT DO
    END IF
  LOOP

END SUB

SUB Evaluate_target
  CALL Explode(line$, field$, "> ")
  
  IF Size(field$) < 1 THEN CAUSE ERROR 31, "Bad line in Find_next_target"
  LET target_name$ = field$(1)
  DO                                 ! Read length (may require 1 or 2 lines)
    CALL InputBlast
    CALL Explode(line$,field$," =")
    IF UCASE$(field$(1)) = "LENGTH" THEN
      LET target_length = Val(DelChar$(field$(2),","))         ! Remove commas from length
      EXIT DO
    END IF
  LOOP

END SUB

!*************************************************************************************************************
!
!	Deals with matches between query and target
!
!*************************************************************************************************************

SUB Find_next_hit
  ! If a target was found, then it's guaranteed that there's at least one hit
  ! Hits begin with " Score"

  LET hit_found = false

  DO WHILE MORE #BlastFile
    CALL InputBlast
    IF line$[1:6] = " Score" THEN 
       LET hit_found = true
       EXIT DO
    END IF
  LOOP

END SUB

SUB Find_seq
  ! If a hit was found, then it's guaranteed that there's at least one line of sequence
  ! In finding the end of the sequence, necessarily either the next hit, target, query,
  !    or end of file will be encountered

  DO WHILE MORE #BlastFile

    IF line$[1:6] = "Query:" THEN		! Next line of sequence alignment
      CALL Evaluate_extent 
    ELSE IF line$[1:8] = " Score =" THEN	! Ran into next hit
      CALL Close_hit
      LET hit_found = true
      EXIT DO
    ELSE IF line$[1:1] = ">" THEN		! Ran into next target
      CALL Close_hit
      LET hit_found = false
      LET target_found = true
      EXIT DO
    ELSE IF line$[1:6] = "Query=" THEN		! Ran into next query
      CALL Close_hit
      LET hit_found = false
      LET target_found = false
      LET query_found = true
      EXIT DO
    ELSE IF line$[1:10] = "Parameters" THEN	! Ran into end of file
      CALL Close_hit
      LET hit_found = false
      LET target_found = false
      LET query_found = false
      EXIT DO
    END IF

    CALL InputBlast

  LOOP

END SUB

!*************************************************************************************************************
!
!	deals with the alignment of the blast output file.  Get the start and end of both query and subject out,
! as well as the sequence of the hit from the subject.
!
!*************************************************************************************************************

SUB Evaluate_extent
  CALL Explode(line$,field$," ")
  LET BeginningOfLine = Val(field$(2))
  LET EndOfLine = Val(field$(4))
  IF begin_query = 0 THEN LET begin_query = BeginningOfLine
  LET end_query = EndOfLine
  CALL InputBlast                    ! Skip alignment marks
  CALL InputBlast                    ! Get Hit sequence
  CALL Explode(line$,field$," ")
  LET BeginningOfLine = Val(field$(2))
  LET EndOfLine = Val(field$(4))
  IF begin_target = 0 THEN LET begin_target = BeginningOfLine
  LET end_target = EndOfLine
END SUB


!*************************************************************************************************************
!
!	Deals with the score section of the blast output file.  Get the e value, percent identity and 
! orientation of the hit out 
!
!*************************************************************************************************************

SUB Evaluate_score
  CALL Explode(line$,field$," ,=")
  IF Size(field$) < 8 THEN CAUSE ERROR 21,"Wrong line in Evaluate_score, line 1"
!  LET e_val = Val(field$(6))
  LET e_val$ = field$(6)
  CALL InputBlast
  CALL Explode(line$,field$," ()=%")
  IF Size(field$) < 9 THEN CAUSE ERROR 21, "Wrong line in Evaluate_score, line 2"
  LET id = Val(field$(3))
  IF field$(9) = "Plus" THEN LET orientation$ = "d" ELSE LET orientation$ = "c"
END SUB

!*************************************************************************************************************
!
!	All information regarding a hit have been read.  Output the information and reset variables so we can 
! move on
!
!*************************************************************************************************************

SUB Close_hit
  IF hit_found = true THEN CALL Output_hit
  LET hit_found = false
  LET begin_query, begin_target = 0
  LET end_query, end_target = 0
  LET hit_seq$ = ""
END SUB


!*************************************************************************************************************
!
!	Prints out all the values we have for the hit in one comma delimited line.
!
!*************************************************************************************************************

SUB Output_hit
  PRINT #output: query_name$; comma$; query_length$; comma$;
  PRINT #output: begin_query; comma$; end_query; comma$;
  PRINT #output: target_name$; comma$; target_length; comma$;
  PRINT #output: orientation$; comma$;
  PRINT #output: begin_target; comma$; end_target; comma$;
  PRINT #output: e_val$; comma$; id
END SUB

END