Using labels in combination with length filter seems to cause problems

[claczny]

When I take a fasta file that contains sequences that are below the length threshold (e.g., 1 knt) and want to use a matching labels file, I get the following error:

2014-09-25 16:56:35,995 DEBUG [TSNERunner] (DataSetUtils.java:318) - TSNE: Fitting performed in 0.00 seconds.
2014-09-25 16:56:36,426 DEBUG [TSNERunner] (DataSetUtils.java:318) - TSNE: Wrote the 60079 x 2 data matrix successfully!
2014-09-25 16:56:36,426 DEBUG [TSNERunner] (DataSetUtils.java:318) - TSNE:
2014-09-25 16:56:36,525 DEBUG [Thread-0] (ProcessInput.java:88) - Points created.
java.lang.IndexOutOfBoundsException: Index: 60079, Size: 60079
    at java.util.ArrayList.rangeCheck(ArrayList.java:638)
    at java.util.ArrayList.get(ArrayList.java:414)
    at lcsb.vizbin.service.DataSetFactory.createDataSetFromPointFile(DataSetFactory.java:75)
    at lu.uni.lcsb.vizbin.ProcessInput$2.run(ProcessInput.java:184)
2014-09-25 16:56:36,759 DEBUG [Thread-0] (ProcessInput.java:88) - Error! Check the logs.
2014-09-25 16:56:36,761 ERROR [Thread-0] (ProcessInput.java:250) - Index: 60079, Size: 60079
java.lang.IndexOutOfBoundsException: Index: 60079, Size: 60079
    at java.util.ArrayList.rangeCheck(ArrayList.java:638)
    at java.util.ArrayList.get(ArrayList.java:414)
    at lcsb.vizbin.service.DataSetFactory.createDataSetFromPointFile(DataSetFactory.java:75)
    at lu.uni.lcsb.vizbin.ProcessInput$2.run(ProcessInput.java:184)

When I do the length filtering before and include a matching labels file, VizBin runs through.
Probably that's a bug in connecting the labels with the sequences where the labels are not properly matched to the length-selected sequences.

[fwhelan]

I might be running into this issue as well:

grep ">" genus_idtxt.fna | wc -l
20282
wc -l genus_idtxt.ann
20283

less confirms that genus_idtxt.ann's first line is label; each following line is the name of the bin that corresponds to the fna file.
When I load fna and ann as the File to visualize and Annotation file, respectively, I do not get an error but the annotation file is ignored (all dots blue). I do have some contigs < 1000. When I restart VizBin and change the minimal contig length to 10, then everything goes through as expected.

[claczny]

Thank you for confirming this 👍

It is easy to workaround this issue by ensuring that the input is all >= 1,000 nt and that the annotation file matches that.
Although the size filtering integrated into VizBin becomes ineffective then.
However, this integrated filtering is IMO oftentimes a very convenient feature ;)
Yet, if the user already uses/creates an annotation file, size-filtering the sequences before using VizBin shouldn't be an issue.
Nevertheless, it is inconvenient and will be corrected in a future release. :)