Hello,

I think your system is running out of memory when processing the full dataset, which is why Velveth is getting killed. Since it works fine with smaller data, your 32GB RAM is likely the bottleneck. You can try a few things to make it work. First, reduce memory usage by lowering the k-mer size in the -shortPaired option (try 21, 27, or 31 instead of something too high) and using -fastq.gz if your data isn’t already compressed. Second, increase your system's swap space since you have a 1TB SSD—creating a 64GB swap file (fallocate -l 64G /swapfile, then enabling it with mkswap and swapon) might help prevent crashes. Third, test it with a smaller subset of reads (e.g., 5-10 million) to see if that runs smoothly. Also, check if your multi-threading settings are using too much memory; reducing the number of threads might help. Lastly, check system logs with dmesg | tail right after the crash—if you see “Killed process due to lack of memory,” then RAM is definitely the issue. If all else fails, you might need to either find a high-memory server or preprocess your data to reduce memory usage before assembly.