FAQ
General¶
Why do I need KGGSum?
KGGSum is fast, convenient, and integrates multiple analysis functions for large-scale GWAS summary data mining.
How do I cite KGGSum?
The publication of the specific method(s) you use should be cited to allow readers to understand the principles of your analyses.
Feedback and Support¶
Where do I report bugs or feedback?
You can send an email to limx54@163.com for this.
Using AI Assistance¶
How can I use generative AI to assist with KGGSum?
You can print the full documents of the online user manual as a PDF file and upload the PDF file to the AI and then ask questions like:
“Please generate commands for me to perform gene-based analysis using KGGSum with my GWAS summary file (
abc.tsv.gz) and reference genome = hg19”.
Troubleshooting¶
Java Version Issue (KGGSum)¶
Symptoms¶
-
When running Association → Region/Gene-based association analysis, you may see:
-
java.lang.UnsupportedClassVersionError: org/apache/arrow/memory/BufferAllocator ... class file version 55.0 ... up to 52.0 - Followed by:
java.lang.ArrayIndexOutOfBoundsException: -1
Root Cause¶
- Apache Arrow integrated in KGGSum v1.0 is compiled with Java 11 (class version 55).
- Your runtime environment only provides Java 8 (class version 52), causing class loading to fail.
- Thread-pool tasks exit unexpectedly, triggering a secondary exception (array out-of-bounds).
Solution¶
- Install OpenJDK 17 on Ubuntu:
sudo apt install openjdk-17-jre - Run KGGSum with the newer Java runtime:
/usr/lib/jvm/java-17-openjdk-amd64/bin/java -jar kggsum.jar - Result: the issue disappears; the Association example command completes normally with full output.
Conclusion¶
- In practice, Java Runtime ≥ 21 is more reliable (even though the manual states ≥ 17).
gsMap Installation Environment Issue¶
Symptoms¶
-
In Conda (Python 3.13), running
pip install gsmapmay fail becausenumpy<2.0.0needs to be built from source, with errors such as: -
ERROR: Unknown compiler(s)... [Errno 2] No such file or directory: 'gcc'
Root Cause¶
- Python 3.13 may not have an available prebuilt NumPy wheel, so pip falls back to building from source.
- Your system does not have a C/C++ compiler installed (gcc/clang), so the build fails.
- Ecosystem support is still catching up: gsmap is not yet well-supported on Python 3.13.
Solution¶
- Create a new Python 3.10 environment:
conda create -n gsMap python=3.10 - Install dependencies in the new environment:
conda install numpy, thenpip install gsmap - Result: gsmap installs and works normally.
Note¶
- TC-04 depends on gsmap. Due to local Python 3.13 and a missing compiler, this testing round does not cover sDESE.