Incomplete edits in alignments

[tomokveld]

Hi, I am encountering some unexpected behavior, it seems that not all bases are accounted for in the edit fields of certain alignments.

I have been aligning onto a single genome "graph", and comparing the results with vg. In most cases the alignments look fine, but in some cases the edits don't cover all the bases of a read. To illustrate consider the (below) 100bp alignments of one read with GraphAligner (92 bp accounted for) and vg (100 bp accounted for). Is this supposed to happen?

I also tried doing the alignments on an actual variation graph, and the bases of this particular read seem to then be properly accounted for in the edit field (although there are other reads which still have missing bases).

Thanks,

Tom

---

GraphAligner

{
  "identity": 0.945054945054945,
  "name": "ERR239432.3455336",
  "path": {
    "mapping": [
      {
        "edit": [
          {
            "from_length": 76,
            "to_length": 76
          },
          {
            "sequence": "C",
            "to_length": 1
          },
          {
            "from_length": 2,
            "to_length": 2
          },
          {
            "sequence": "TT",
            "to_length": 2
          },
          {
            "from_length": 2,
            "to_length": 2
          },
          {
            "sequence": "GG",
            "to_length": 2
          },
          {
            "from_length": 7,
            "to_length": 7
          }
        ],
        "position": {
          "is_reverse": true,
          "node_id": "40236",
          "offset": "181"
        }
      }
    ]
  },
  "score": 13,
  "sequence": "CCCTGCAGCTGAGGCATCTGCTCTCTTTCAACATACTCTCTTCTGGAAAGAGGGAGAGCAGGAAGCATCATGGGTTTCACTTTACGGTTGGGGAAGCTGG"
}  

vg

{
  "identity": 0.76,
  "mapping_quality": 60,
  "name": "ERR239432.3455336",
  "path": {
    "mapping": [
      {
        "edit": [
          {
            "from_length": 76,
            "to_length": 76
          },
          {
            "sequence": "TCACTTTACGGTTGGGGAAGCTGG",
            "to_length": 24
          }
        ],
        "position": {
          "is_reverse": true,
          "node_id": "40236",
          "offset": "181"
        },
        "rank": "1"
      }
    ]
  },
  "quality": "Ix8gHx4mICQiGCImICYWJR8fJh8gICMnIR8mIh8nHhwhJhkeHSEdJCYkIyAmIxwbHB8nHRodDxEdDx0ZIRwMIiUgJCMdICUiJiImICYhGSImJScmJCMnIiUZFhklJSYeBwYXFg==",
  "refpos": [
    {
      "is_reverse": true,
      "name": "6",
      "offset": "40219069"
    }
  ],
  "score": 81,
  "sequence": "CCCTGCAGCTGAGGCATCTGCTCTCTTTCAACATACTCTCTTCTGGAAAGAGGGAGAGCAGGAAGCATCATGGGTTTCACTTTACGGTTGGGGAAGCTGG",
  "time_used": 1345
}

[maickrau]

Hi, could you upload the read and the graph as well?

[tomokveld]

Yes, sure. It's too big for github, so I put it on Dropbox: https://www.dropbox.com/sh/zw8dc4z1cbpz03y/AABrPTzxSeGQu1GkZvu0iHCBa?dl=0

[maickrau]

Fixed in commit 8e37ecb. Will be included in the next bioconda release.

The output edit fields were missing indels in some cases, here the first 8bp of a 9bp insertion were missing. The identity field also reported too high identities because of this.

[tomokveld]

Thanks, I will have a look at it.