Fatal protobuf error when providing seed file (.gam)

[IgnazioSpaccavento]

Hello,
I'm encountering a protobuf fatal error when trying to align using a seed file (.gam).
GraphAligner crashes immediately after loading the seeds.

Command used:
./bin/GraphAligner
-g ../res/test/C4/graph/e_C490.gfa
-f /res/chains.fq
-s /res/seeds.gam
-a /res/test/graphaligner_output.gam
-b 35

Output / Error:
GraphAligner Branch master commit 3c048f2 2025-07-25 14:29:37 +0300
Load graph from ../res/test/C4/graph/e_C490.gfa
Build alignment graph
Load seeds from /res/seeds.gam
[libprotobuf FATAL /home/ignazio/miniconda3/envs/GraphAligner/include/google/protobuf/repeated_ptr_field.h:271] CHECK failed: (index) < (current_size_):
terminate called after throwing an instance of 'google::protobuf::FatalException'
what(): CHECK failed: (index) < (current_size_)
Aborted (core dumped)

Environment

• GraphAligner commit: 3c048f2 (master)
• OS: (Ubuntu 22.04)
• Installation: compiled from source (conda env)

Could this be caused by a mismatch between GraphAligner’s expected GAM schema and the one generated by vg?
Thanks in advance for any help!

[maickrau]

Could you please share the seed file, the graph file and the input sequence file?

[IgnazioSpaccavento]

I'm attaching a zip file with the requested files.
Thank you very much for your time.

seed.zip

[maickrau]

Looks like the gam seed input parameter -s has several bugs in it. You can instead use the parameter --realign which instead uses seed input in gaf format.

Also, looks like a couple of the seeds in node_id 1 have offsets which are outside the node sequence. You can try filtering out those seeds and aligning with these commands:

vg view -aj seeds.gam > seeds.json
grep -vP "\"offset\": \"5[0-9]{4}\"" < seeds.json >seeds_no50k.json
vg view -aJ -G seeds_no50k.json > seed_no50k.gam
vg convert -gv graph.gfa > graph.vg
vg convert -G seed_no50k.gam graph.vg > seeds_no50k.gaf
awk '$6!="*"' < seeds_no50k.gaf > seeds_no50k_nounaligned.gaf

GraphAligner -g graph.gfa -f read.fq --realign seeds_no50k_nounaligned.gaf -a aln.gaf -b 35

The lines above the GraphAligner call are just to filter out strange seeds, either those which are located outside a node's sequence or those which are for some reason marked as unaligned sequences in the gaf file. In general you can just make a gaf file and input that with the parameter --realign.

I'll leave this issue open until I've fixed the bugs with -s.