Gene loss in subsequent round of maker for fungal genome annotation

Question

[urmi208]

Hello,

I am trying to run maker2.31.9 on a fungal genome. I have got transcript and protein sequences from a related species using parameters "altest" and "protein" in the maker_opts.ctl file. Here is the protocol I am using:

1. Run maker with repeat masking and providing transcript and protein sequences from related species (Run A)
2. Create SNAP model with CEGMA
3. Train Augustus with BUSCO
4. Run (run B ) with the new SNAP (done at step 2) and augustus species with options turned off (est2genome=0) and (protein2genome=0) data, provide gff file (altest_gff=runA_cdna2genome.gff, protein_gff=runA_protein2genome.gff3)
5. Create SNAP model from run B.
6. Train Augustus with transcripts from run B and BUSCO
7. Run (run C ) with the new SNAP (done at step 5) and augustus species with options turned off (est2genome=0) and (protein2genome=0) data, provide gff file (altest_gff=runA_cdna2genome.gff, protein_gff=runA_protein2genome.gff3), keep_preds=1

As a result of this, I get following gene numbers:

• run A: 12532 total genes out of which 12490 have AED < 0.5
• run B:10038 total genes out of which 9996 have AED < 0.5
• run C: 12909 total genes out of which 12133 have AED < 0.5

I don't understand why there is gene loss for run B. Please could anyone shed come light on this?

Many thanks in advance.