Details
-
Type:
Task
-
Status: Closed (View Workflow)
-
Priority:
Major
-
Resolution: Done
-
Affects Version/s: None
-
Fix Version/s: None
-
Labels:None
-
Story Points:4
-
Epic Link:
-
Sprint:Fall 1, Spring 3, Spring 5, Spring 6, Spring 8, Summer 2, Summer 3, Summer 4, Summer 6, Fall 2, Fall 3, Fall 4, Fall 5
Description
Identified two related issues in IGB 10.2.0 involving the loading of VCF files, both of which are regressions from version 10.1.0. These issues negatively affect performance and functionality when working with large or indexed VCF datasets.
1. Performance and Memory Issue When Loading VCF Files
When loading larger VCF files (e.g., 1KG.chr22.anno.infocol.vcf.gz), IGB 10.2.0 exhibits significantly increased memory usage and may crash or freeze when navigating to a gene.
In IGB 10.1.0, loading this file and accessing gene-level data works as expected.
In IGB 10.2.0, it appears the entire file is being loaded into memory when "Load Data" is clicked, rather than just the in-frame genomic region (as was the behavior in 10.1.0).
This results in performance degradation and potential out-of-memory errors, especially with large datasets.
2.* Failure to Load Tabix-Indexed VCF Files*
When attempting to load a VCF file with an associated Tabix index (e.g., Genome in a Bottle VCF), IGB 10.2.0 throws the following error:
ClassCastException: VCFSymLoaderTabix cannot be cast to QuickLoadSymLoader
These same files load without issue in IGB 10.1.0.
This appears to be a class loading or module registration issue introduced in the newer version, likely related to recent changes in VCF parsing logic.
Attachments
Issue Links
Activity
Branch changes: https://bitbucket.org/lorainelab-deepthi/integrated-genome-browser/branch/IGBF-4106
Testing guide:
Please refer to Ticket IGBF-4219 for specific files to test and issues to address while testing.
Downloaded the early release version of IGB from bioviz.org and tested on my Mac. VCF files with space-separated headings are being handled with appropriate warning messages. VCF files with an index in the same directory are loading much faster than they do without an index. VCF files can be added via URL or locally and seem to be working consistently.
Closing ticket!