vidjil-to-fasta.py: Percentage on total of reads

This will allow to more reliably add percentages that otherwise would give
meaningless results.

In the test we now have 497 reads out of 786861.

tools/tests/should-get-tests/vidjil-to-fasta-with-sequences.should-get

@@ -7,4 +7,4 @@ $ Affectation for sequence YYGGGYYACGYAYAGCGGYGYTTYYCCTYTYTGYTYTGCYAAAYAACYYYYTG
 1:VDJ 1 95 100 105 107 134
 
 $ Percentage for sequence GGAGTCGGGGGAGGCTTGGTACAGCCTGGCAGGTCCCTGAGACTCTCCTGTGCAGCCTCTGGATTCACCTTTGATGATTATGCCATGCACTGGGTCCGGCAAGCTCCAGGGAAGGGCCTGGAGTGGGTCTCAGGTATTAGTTGGAATAGTGGTAGCATAGGCTATGCGGACTCTGTGAAGGGCCGATTCACCATCTCCAGAGACAACGCCAAGAACTCCCTGTATCTGCAAATGAACAGTCTGAGAGCTGAGGACACGGCCTTGTATTACTGTGCACCCGGAGGTATGGACGTCTGGGGCCAAGGGACCCTGGTCACC
-1:IGHV3-9.01 7/CCCGGA/17 IGHJ6.02 .* percentage=0.304534313725
+1:IGHV3-9.01 7/CCCGGA/17 IGHJ6.02 .* percentage=0.00066947

tools/vidjil-to-fasta.py

@@ -215,8 +215,7 @@ def write_fuse_to_fasta(data, outfile, used_names, current_filename, options, me
             max_sample = max(clone.d['reads'])
             #take the index corresponding to the max_sample
             index_max_sample = clone.d['reads'].index(max_sample)
-            germline = clone.d['germline']
-            reads_total_nb = data.d['reads'].d['germline'][germline][index_max_sample]
+            reads_total_nb = data.d['reads'].d['segmented'][index_max_sample]
             percentage = float(max_sample)/reads_total_nb
 
             if name in used_names: