From 1adfa9084a13ff96b34ea614e024148a172af7b9 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Sat, 11 Oct 2025 18:15:27 +0200
Subject: On pangenotypes

---
 topics/genetics/pangenotypes.gmi | 46 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)
 create mode 100644 topics/genetics/pangenotypes.gmi

diff --git a/topics/genetics/pangenotypes.gmi b/topics/genetics/pangenotypes.gmi
new file mode 100644
index 0000000..9ef1c8e
--- /dev/null
+++ b/topics/genetics/pangenotypes.gmi
@@ -0,0 +1,46 @@
+# Pangenotypes
+
+## GRG format
+
+
+Looking for graph genotyping I ran into Genotype Representation Graphs GRG
+
+=> https://pmc.ncbi.nlm.nih.gov/articles/PMC11071416/
+
+It has a binary storage format that represents something like:
+
+```
+# GRG file example: genotype graph
+# Nodes section: NODE <id> <label> allele=<genotype>
+NODE 1 GeneA allele=AA
+NODE 2 GeneB allele=AG
+NODE 3 GeneC allele=GG
+NODE 4 GeneD allele=AA
+NODE 5 GeneE allele=AG
+
+# Edges section: EDGE <from_id> <to_id>
+EDGE 1 2
+EDGE 1 3
+EDGE 2 4
+EDGE 3 4
+EDGE 4 5
+EDGE 5 1
+```
+
+the tooling
+
+=> https://github.com/aprilweilab/grgl.git
+
+builds with
+
+```
+guix shell -C -N coreutils gcc-toolchain make cmake openssl nss-certs git pkg-config zlib
+```
+
+I did some tests and read the source code. The nice thing is that they have very similar ideas. Unfortunately the implementation is not what we want. I wonder why people alway reinvent data structures :/. To get an idea:
+
+=> https://github.com/aprilweilab/grgl/blob/main/src/serialize.cpp
+
+I would like to take similar ideas and take it to an efficient in-memory graph structure that is easily extensible. RDF is key for extensions (and queries). A fast RDF implementation we are going to try is
+
+=> https://pyoxigraph.readthedocs.io/en/stable/index.html
-- 
cgit 1.4.1