PERL subroutines and references Andrew Emerson High Performance

PERL subroutines and references Andrew Emerson, High Performance Systems, CINECA

Consider the following code: # Counts Gs in various bits of DNA $dna=“CGGTAATTCCTGCA”; $G_count=0; for ($pos=0; $pos <length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$G_count if ($base eq ‘G’); } # end for. . . do something else $new_dna = <DNA_FILE>; $G_count=0; for ($pos=0; $pos <length $new_dna; $pos++) { $base=substr($new_dna, $pos, 1); ++$G_count if ($base eq ‘G’); } # end for

Inconvenient to repeat pieces of code many times if it does the same thing. Better if we could write something like… # Counts Gs in various bits of DNA # improved version (PSEUDO PERL) # Main program $dna=“CGGTAATTCCTGCA”; count_g using $dna; . . do something else $new_dna = <DNA_FILE>; count_g using $new_dna; . . count_G subroutine for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$G_count if ($base eq ‘G’); }

Subroutines Ø The pieces of code used in this way are often called subroutines and are common to all programming languages (but with different names) Ø subroutines (Perl, FORTRAN) Ø functions (C, C++*, FORTRAN, Java*) Ø procedures (PASCAL) Ø Essential for procedural or structured programming. * object-oriented programming languages

Advantages of using subroutines Ø Saves typing → fewer lines of code →less likely to make a mistake Ø re-usable Øif subroutine needs to be modified, can be changed in only one place Øother programs can use the same subroutine Øcan be tested separately Ø makes the overall structure of the program clearer

Program design using subroutines Conceptual flow subroutines can use other subroutines to make more complex and flexible programs

Program design using subroutines -pseudo code # # Main program # pseudo-code. . set variables. call sub 1. call sub 2. call sub 3. exit program sub 1 # code for sub 1 exit subroutine sub 2 # code for sub 1 exit subroutine sub 3 # code for sub 1 call sub 4 exit subroutine sub 4 # code sub 4 exit

Using subroutines in Perl Example 1. # Program to count Gs in DNA sequences # (valid perl) # Main program $dna=“GGCCTAACCTCCGGT”; count_G; print “no. of G in $dna=$number_of_gn”; # subroutines sub count_G { for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$number_of_g if ($base eq ‘G’); } # end for }

Subroutines in Perl Defined using the sub command: sub name {. . . } Called from the main program or another subroutine using its name: name; Sometimes you will see in old Perl programs (like mine) &name; But is optional in modern Perl.

Subroutines in Perl Subroutines can be placed anywhere in the program but best to group them at the end; # main program. . exit; # subroutines defined here sub 1 { . . . } sub 2 { . . . } sub 3 { . . . } exit not strictly necessary, but makes it clear we want to leave the program here.

Return to example 1 why is this bad? # Program to count Gs in DNA sequences # Main program What does count_G need ? $dna=“GGCCTAACCTCCGGT”; count_G; print “no. of G in $dna=$number_of_gn”; exit; # subroutines sub count_G { for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$number_of_g if ($base eq ‘G’); } # end for } Where did this come from?

Perl subroutines-passing parameters The input/outputs to a subroutine are specified using parameters (or arguments) given when the subroutine is called: $no_of_G = count_g($dna); It is now clear what the subroutine expects as input and what is returned as output. Other examples: $day_in_yr = calc_day($day, $month); $num_sequences = read_sequence_file(@database);

Input parameters All the parameters in a Perl subroutine (including arrays) end up in a single array called @_ Therefore in the code: # $pos = find_motif($motif, $protein); . sub find_motif { $a takes the value of $motif $a = $_[0]; $b = $_[1]; . . . } $b takes the value of $protein

Subroutine output (return values) A subroutine does not have to explicitly return something to the main program: print_title; sub print_title{ print “Sequence Manipulation programn”; print “---------------n”; print “Written by: A. Nother n”; print “Version 1. 1: n” } but often it does, even if only to signal the procedure went well or gave an error.

Subroutine return values By default the subroutine returns the last thing evaluated but you can use the return statement to make this explicit: input sub count_G { $dna=@_[0]; for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$number_of_g if ($base eq ‘G’); } # end for return $number_of_g; } output return also exits the subroutine

Return Values You can return from more than 1 point in the sub – can be useful for signalling errors: if ($dna eq “”) { return “No DNA was given”; # exit with error message } else {. . } # end if return $number_of_G; } # end sub In which case the calling program or sub should check the return value before continuing. However, in general best to return from only 1 point, otherwise difficult to follow the logic.

Return values Can also return multiple scalars, arrays, etc. but just as for the input everything ends up in a single array or list: @DNA = read_file($filename); . . ($n. G, $n. C, $n. T, $n. A) = count_bases($dna); sub count_bases {. . . return ($num_G, $num_C, $num_T, $num_A); } # end sub note ( and ) for the list

Counting bases – Attempt 2 # Program to count Gs in DNA sequences # using input/output parameters # Main program better but. . . $dna=“GGCCTAACCTCCGGT”; $num_g = count_G($dna); print “no. of G in $dna=$num_gn”; exit; # subroutines sub count_G { $dna=$_[0]; All the variables inside are also visible outside the sub, not only params ! for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); ++$number_of_g if ($base eq ‘G’); } # end for return $number_of_g; # return value of sub

The need for variable scoping A subroutine written this way is in danger of overwriting a variable used elsewhere in the program. Remember that a subroutine should work like a black box, apart from welldefined inputs/outputs it should not affect the rest of the program. input Apart from input/output all vars needed by the sub should appear and disappear within the sub. Allows us also to use the same names for vars outside and inside the sub without conflict. sub output

Variable scoping in Perl Ø By default, all variables defined outside a sub are visible within it and vice-versa – all variables are global. Ø Therefore the sub can change variables used outside the sub. Solution ? Restrict the scope of the variables by making them local to the subroutine → eliminate the risk of altering a variable present outside the sub. Also makes it clear what the subroutine needs to function. .

Variable scoping in Perl In Perl, variables are made local to a subroutine (or a block) using the my keyword. For example, my variable 1; # simple declaration my $dna=“GGTTCACCACCTG”; # with initialization my ($seq 1, $seq 2, $seq 3); # more than 1 Attenzione my $seq 1, $seq 2; Must use () if multiple vars per line This means my $seq 1; $seq 2; Which is valid Perl so the compiler won’t give an error.

Subroutines with local variables # Program to count Gs in DNA sequences – final version # Main program $dna=“GGCCTAACCTCCGGT”; $num_g = count_G($dna); print “no. of G in $dna=$num_gn”; Remember that my exit; makes a copy of the # subroutines variable. sub count_G { my $dna=$_[0]; my ($pos, $base); my $number_of_g=0; for ($pos=0; $pos<length $dna; $pos++) { $base=substr($dna, $pos, 1); Variables declared like ++$number_of_g if ($base eq ‘G’); this is called lexical scoping in the Perl man } # end for pages return $number_of_g; # return value }

$Other examples of subroutine use sub find_motif { my $motif=shift; my $protein=shift; . .$

Other examples of subroutine use sub find_motif { my $motif=shift; my $protein=shift; . . . } # shifts the @_ array # (avoids $_[0], etc) sub count_C_and_G { my @fasta_lines=@_; . . . return ($num_C, $num_G); } sub reverse_seq { use strict; $seq=$_[0]; . . . } # returns a list # the strict command enforces use of my # so this line will give an error (even # if defined in main program)

Question: What if we want to pass two arrays ? $seq=compare_seqs(@seqs 1, @seqs 2); Remember that everything arrives in the sub in a single array. Likewise for return values: (@annotations, @dna) = parse_genbank(@dna); . . . sub parse_genbank {. . . return (@annotations, @dna); } In the first example what will be in the special array @_ ? Solution: Use references

subroutines – call by value Consider $i=2; $j =add_100($i); print “i=$in”; $i unaffected by the subroutine sub simple_sub { my $i=$_[0]; $i=$i+100; return $i; } This is called Call by Value because a copy is made of the parameter passed and whatever happens to this copy in the subroutine doesn’t affect the variable in the main program

What are references? References can be considered to be an identifier or some other description of the objects rather than the objects themselves. To buy Bananas Beer Fruit Pasta Frozen pizza more beer reference or copy

References In computing, references are often addresses of objects (scalars, arrays, . . ) in memory: @genbank $dna 100 address of $dna scalar 200 address of @genbank array 300

References Ø References (sometimes called pointers in other languages) can be more convenient because in Perl they are scalars → often much smaller than the object they refer to (e. g. an array or hash). Ø Array references can be passed around and copied very efficiently, often also using less memory. Ø Being scalars, they can be used to make complicated data structures such as arrays of arrays, arrays of hashes and so on. .

References in Perl Simplest way to create a reference in Perl is with $scalar_ref = $sequence; # reference to a scalar $dna_ref = @DNA_list; # reference to an array $hash_ref = %genetic_code; # reference to a hash To get back the original object the reference needs to be dereferenced; $scalar = $$scalar_ref; # for scalars just add $ @new_dna = @$dna_ref; # for arrays just add @ %codon_lookup = %$hash_ref; # similary for hashes

Passing two arrays into a sub using references # compare two databases, each held as an array # $results = compare_dbase(@dbase 1, @dbase 2); # supply refs. . . sub compare_dbase { my ($db 1_ref, $db 2_ref) = @_; # params are refs to arrays @db 1 = @$db 1_ref; # dereference @db 2 = @$db 2_ref; # dereference. . . # now use @db 1, @db 2 return $results; } Similarly we can return 2 or more arrays by the same method

References – final words Caution: Calling by reference can change the original variables; OUTPUT @dna 1=(G, G, T, C, T, G); dna 1=G G T C T G A A A @dna 2=(A, A, A); dna 2=A A A add_seqs(@dna 1, @dna 2); print “dna 1=@dna 1 n dna 2=@dna 2 n”; sub add_seqs { my ($seq 1, $seq 2) =@_; push(@$seq 1, $@seq 2); } If you don’t want this behaviour then create local copies of the arrays as in previous example.

subroutines-summary Ø subroutines defined with sub represent the main tool for structuring programs in Perl. Ø variables used only by the subroutine should be declared with my, to prevent conflict with external variables (lexical scoping) Ø parameters passed in to the sub end up in the single array @_; similarly for any return values Ø array references need to be used to pass two or more arrays in (call by reference) or out of a sub.