Towards a 2D RG-PG-SGD

[subwaystation]

echo "gi|345525392:5000-18402" > ref.txt
odgi sort -i DRB1-3123_unsorted.og -o DRB1-3123.YH.og -H ref.txt -Y
odgi viz -i DRB1-3123.YH.og -o DRB1-3123.YH.og.png -du
odgi view -i DRB1-3123.YH.og -g > DRB1-3123.YH.gfa
odgi layout -i DRB1-3123.YH.gfa -o DRB1-3123.YH.gfa.layH -T DRB1-3123.YH.gfa.layH.tsv -P -t 1 -H ref.txt
./target/release/gfaestus ~/Downloads/ref/lay/DRB1-3123.YH.gfa ~/Downloads/ref/lay/DRB1-3123.YH.gfa.layH.tsv

This gives me

[AndreaGuarracino]

It seems hot

---

From: Simon Heumos @.> Sent: 29 August 2023 15:53 To: pangenome/odgi @.> Cc: Subscribed @.***> Subject: Re: [pangenome/odgi] Towards a 2D RG-PG-SGD (PR #491)

echo "gi|345525392:5000-18402" > ref.txt odgi sort -i DRB1-3123_unsorted.og -o DRB1-3123.YH.og -H ref.txt -Y odgi viz -i DRB1-3123.YH.og -o DRB1-3123.YH.og.png -du odgi view -i DRB1-3123.YH.og -g > DRB1-3123.YH.gfa odgi layout -i DRB1-3123.YH.gfa -o DRB1-3123.YH.gfa.layH -T DRB1-3123.YH.gfa.layH.tsv -P -t 1 -H ref.txt ./target/release/gfaestus ~/Downloads/ref/lay/DRB1-3123.YH.gfa ~/Downloads/ref/lay/DRB1-3123.YH.gfa.layH.tsv

This gives me

[image]https://user-images.githubusercontent.com/10184961/264042450-6de40ea9-d9c0-40a8-a3f0-6efdb3181617.png

— Reply to this email directly, view it on GitHubhttps://github.com/pangenome/odgi/pull/491#issuecomment-1697491431, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AO26XHW2RAKBLLC5HIYCGETXXXX6ZANCNFSM6AAAAAAWRRUMXM. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[subwaystation]

aaah, oh god 🤥🫠

my implementation works as I intended, but conceptually, it is ***** in my naivity, I am just pasting the reference sequences from left to right so if the 2nd reference sequence has nodes shared with the first one, I would generate long links ☠️ --> my implementation works well for one single reference genome or for one haplotype that is distributed across contigs in the graph, but not for several reference genomes🤌

I see 2 ways:

• I take all reference sequences, fix their nodes and apply PG-SGD
• Progressive approach as explained by erikg I take the 1st reference, paste it from left to right, fix its nodes and add the 2nd reference sequence by apply PG-SGD only to the unfixed nodes of the 2nd reference. In the next iteration, only the nodes of the 3rd reference sequence are unfixed, .....
• Are there other ideas?