It depends on what you are trying to achieve. Generally speaking, E-value will be more informative than the parameters you listed. If you are looking for orthologs, E <= 1e^-20 is probably a good starting point, along with high coverage. If you want to identify paralogs, or in general more distantly related proteins, E <= 1e^-5 should be a good cutoff. In the latter case you are not necessarily expecting either high identity or high coverage.

• https://academic.oup.com/nar/article/47/D1/D807/5160989
• https://www.nature.com/articles/nmeth.3830
• https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-4-35