Table 4.

Summary of recommendations for updates to PP3 and BP4 criteria in the ACMG/AMP recommendations and comparison with the 2015 ACMG/AMP recommendations

Guidance	2015 ACMG/AMP recommendations	Our recommendation
Evidential strength of in silico tools	supporting	PP3_Strong, PP3_Moderate, PP3_Supporting, Indeterminate, BP4_Supporting, BP4_Moderate, and BP4_Strong
Tools to be used	no specific recommendation, although a list of tools provided with the caveat that they have not been validated	tools that reach at least strong evidence for pathogenicity and moderate for benignity; it is valid to use any tool at the indicated evidential strength level as determined by our approach, so long as it is chosen before seeing its scores and preferably any other evidence.
Score thresholds for tools	no explicit recommendations but often defaults to developer-recommended thresholds	specific score thresholds calibrated to different evidential strengths
Number of tools to be used	multiple lines of evidence, often interpreted as multiple tools (no number provided)	combinations of tools was not systematically assessed but results suggest the use of a single tool rather than an uncalibrated consensus of multiple tools
Decision-making	if multiple lines of evidence agree, evidence can be counted as supporting; otherwise, evidence from tools cannot be used	evidence can be counted at the appropriate strength depending on the tool and score threshold chosen; the maximal strength of a tool must be committed to, i.e., if a tool provides at most PP3_Moderate strength and a variant satisfies the corresponding score threshold, it can only be counted as moderate evidence
Combing rules with other evidence codes	PP3 and BP4 can be applied for supporting evidence without consideration of other codes used	limit the combined evidence from PM1 and PP3 to strong