I find XML::Simple to be very simple if the XML is very simple, but with non-trivial XML I am often surprised by what it produces and, if I persist with it, end up spending a lot of time re-reading the documentation and studying Data::Dumper dumps.

Here is an example which shows, step by step as I worked it out, how to access the source_name element...

use strict;
use warnings;
use Data::Dumper;

my $VAR1 = {
    'sas_residue_annotation' => [
        {
            'xmlns'   => 'http://www.ebi.ac.uk/WSsas/Schema',
            'sources' => [
                {
                    'source' => [
                        {
                            'source_name'  => [ '1iho' ],
                            'ref_evalue'   => [ '5.5e-50' ],
                            'ref_overlap'  => [ '282' ],
                            'ref_identity' => [ '47.20' ],
                            'ref_pmid'     => [ '11377204' ]
                        },
                        {
                            'source_name'  => [ '1mop' ],
                            'ref_evalue'   => [ '8.3e-38' ],
                            'ref_overlap'  => [ '264' ],
                            'ref_identity' => [ '43.60' ],
                            'ref_pmid'     => [ '12717031' ]
                        },
                        {
                            'source_name'  => [ '1n2b' ],
                            'ref_evalue'   => [ '8.2e-38' ],
                            'ref_overlap'  => [ '264' ],
                            'ref_identity' => [ '43.60' ],
                            'ref_pmid'     => [ '12717031' ]
                        },
                    ],
                },
            ],
        },
    ],
};

print Dumper($VAR1);
print Dumper($VAR1->{sas_residue_annotation});
print Dumper($VAR1->{sas_residue_annotation}[0]);
print Dumper($VAR1->{sas_residue_annotation}[0]{sources});
print Dumper($VAR1->{sas_residue_annotation}[0]{sources}[0]);
print Dumper($VAR1->{sas_residue_annotation}[0]{sources}[0]{source});

foreach my $source (@{$VAR1->{sas_residue_annotation}[0]{sources}[0]{s
+ource}}) {
    print "source name: ", join(',',@{$source->{source_name}}), "\n";
}
[download]

Note that I wrote each successive print after studying the output of the previous one. It is a bit tedious, but for deeply nested structures I find it faster than trying to do it all in my head and then debugging my errors.