As the data you wish to compare is the accumulated (testtype, result & resultval) and the information you wish to extract are the sampleID's and subID's, possibly the best method would be to build a hash of array's of arrays (HoAoA) as you read your data in. The key would be the accumulated test data, and the array of arrays would contain one array of sampleID's and one of subID's. Doing it this way, there is no searching to be done once the file has been read in and the structure built. For your sample data this might look like this

%data = {
    '17  1  12.3  17  2  9  18  1  17.2' => [ [ 12345, 67890], [ 543D3
+, 775G2 ] ],
    '17  1  12.3  17  2  9'              => [ [ 45678 ], [ 543D3 ] ],
}
[download]

The values for your report can then be read out of the nested arrays directly with no further searching, sorting or matching.

Assuming that your input file is as you have shown it: sorted by sampleID and each set of values follow in a consistant order, building this data structure is a simple one-pass linear affair.

1. open file
2. Initialise $prevID, $results, @sampleIDs, @subIDs, %data;
3. Read a line from the file <FILE>, while lines to read
4. split into $sampleID, $subID, @rest
5. If the $sampleID matches the previous, join the latest set of results to the previous and push the two id's into their respective arrays. Goto next line;
6. else Add an entry to the hash using the accumulated $results string as the key and pushing an anonymous array containing

   [ [@sampleIDs], [@subIDs] ].

   Set $prevID = $sampleID; @sampleIDs = $sampleID; @subIDs = $subID; $results = $rest;

7. loop;
8. close file;
9. for each of the values in %data;
10. @{$_->[0]} is the array of sampleID's that had this set of matching results
11. @{$_->[1]} is the array of subID's that had these results.

Hopefully that pseudo code and the other answers will give you enough clues to get you going. If you get stuck, come back with what you have and someone will nudge you along:)