examples.inc

<?php

global $examples, $example_options;

if (!function_exists('t')) {
	function t($str) { return $str; }
}

# ------------------------------------------------------------------------------------ #

$example_names['q1'] = t('What are proteins in the NR database that have taxonomic name "Escherichia coli"?');
$example_jobs['q1'] = 1;
$examples['q1'] = <<<BOA
s: Sequence = input;
count : output sum[string] of int;

foreach(i:int; def(s.annotation[i]))
    if (strfind( "Escherichia coli", s.annotation[i].tax_name) > -1)
        count[s.seqid]  << 1;
BOA;

# ------------------------------------------------------------------------------------ #

$example_names['q2'] = t('What are the list of conserved proteins?');
$example_jobs['q2'] = 2;
$examples['q2'] = <<<BOA
s: Sequence = input;
protOut : output sum [string][string] of int;

distinctTax := function (seq: Sequence): int{
    taxSet : set of string;
    foreach(i:int; def(seq.annotation[i]))
        add(taxSet,seq.annotation[i].tax_id);
    return(len(taxSet));
};

# we define conserved proteins as those who have > 10 distinct taxonomic assignments
if (distinctTax(s) > 10){
 foreach(i:int; def(s.annotation[i])){
   if (strfind("[",s.annotation[i].defline)> 0)
        protOut [trim(substring(s.annotation[i].defline, 0, strfind("[",s.annotation[i].defline)))][s.seqid] << 1;
   else
        protOut [s.annotation[i].defline][s.seqid]<<1;
  }
}
BOA;

# ------------------------------------------------------------------------------------ #

$example_names['q3'] = t('What are protein sequences and the frequencies of coronavirus in taxonomic assignments?');
$example_jobs['q3'] = 3;
$examples['q3'] = <<<BOA
s: Sequence = input;
count : output sum[string][string] of int;

foreach(i:int; def(s.annotation[i]))
    if (strfind( "coronavirus", s.annotation[i].tax_name) > -1)
        count[s.seqid][s.annotation[i].tax_name]  << 1;
BOA;

# ------------------------------------------------------------------------------------ #

$example_names['q4'] = t('What is the frequency of protein length in the NR database?');
$example_jobs['q4'] = 4;
$examples['q4'] = <<<BOA
s: Sequence = input;
counts: output sum[int] of int;

foreach(i:int; def(s.cluster[i]))
    if (s.cluster[i].similarity==95)
        counts [s.cluster[i].length] <<  1;
BOA;

# ------------------------------------------------------------------------------------ #

$example_names['q5'] = t('What are all clusters belong to a specific protein function SCN?');
$example_jobs['q5'] = 5;
$examples['q5'] = <<<BOA
s: Sequence = input;
counts_protein: output collection[string][string] of int;

foreach(i:int; def(s.cluster[i]))
  if (s.cluster[i].similarity==95)
    if (strfind( "SCN", s.annotation[i].defline) > -1)
	   counts_protein [s.seqid][s.cluster[i].cid] << 1;
BOA;

# ------------------------------------------------------------------------------------ #

$example_names['q6'] = t('What are number of protiens in each phylum in the tree of life?');
$example_jobs['q6'] = 6;
$examples['q6'] = <<<BOA
# search for Streptococcus
s: Sequence = input;
phylCount: output sum [string] of int;
taxs := {"Firmicutes", "Fusobacteria"};

for (j := 0; j < len(taxs); j++)
	exists(i: int; match(taxs[j], s.annotation[i].tax_name))
	 	 phylCount [taxs[j]] << 1;
BOA;
# ------------------------------------------------------------------------------------ #

$example_names['q7'] = t('What are the list of taxonomic assignment for all proteins in NR?');
$example_jobs['q7'] = 31;
$examples['q7'] = <<<BOA
s: Sequence = input;
clstrOut : output collection [string] of string;

getTaxList := function(seq: Sequence):string {
    taxids :="";
    foreach(i:int; def(s.annotation[i])){
        if (s.annotation[i].tax_name !="")
            taxids = taxids + s.annotation[i].tax_id + " ";}
    return taxids ;
};

foreach(i:int; def(s.cluster[i]))
    if (s.cluster[i].similarity==95 && s.cluster[i].representative)
        clstrOut [s.seqid] <<  getTaxList(s);
BOA;


# ------------------------------------------------------------------------------------ #

$example_names['q8'] = t('How many proteins in NR do not have a taxonomic assignment?');
$example_jobs['q8'] = 8;
$examples['q8'] = <<<BOA
s: Sequence = input;
countNull: output sum of int;

foreach (i: int; def(s.annotation[i]))
    if (s.annotation[i].tax_name == "")
    	countNull << 1;
BOA;

# ------------------------------------------------------------------------------------ #

$example_options = array(
	 0 => t('-- Select Example --'),
	 'q1' => $example_names['q1'],
	 'q2' => $example_names['q2'],
	 'q3' => $example_names['q3'],
	 'q4' => $example_names['q4'],
	 'q5' => $example_names['q5'],
	 'q6' => $example_names['q6'],
	 'q7' => $example_names['q7'],
	 'q8' => $example_names['q8'],
);

?>