Psiblast

# 2) Save an ASCII PSSM for inspection or downstream tools
psiblast \
  -query query.fa \
  -db prot_db \
  -num_iterations 3 \
  -out_ascii_pssm query.ascii.pssm

# 3) Restart from an existing checkpoint
psiblast \
  -in_pssm query.pssm \
  -db prot_db \
  -num_iterations 2 \
  -outfmt 6

Recommended Workflow

1. Start from a high-confidence protein query or curated alignment.
2. Set -inclusion_ethresh conservatively so poor hits do not contaminate the profile.
3. Inspect each iteration's accepted hits before trusting convergence.
4. Save the final PSSM if the profile will be reused or compared later.

Guardrails

• Input must be protein sequence, protein MSA, or protein PSSM, never nucleotide query sequence.
• More iterations are not always better; false positives can poison the profile early.
• -evalue controls reporting, while -inclusion_ethresh controls what enters the next-round model.
• Save checkpoints (-out_pssm, -out_ascii_pssm) if the iterative result matters scientifically.