During powder manufacture for metal additive manufacturing, the final composition will deviate from the designed composition, which may lead to undesired properties in the printed part. It is critical to perform an uncertainty quantification during alloy design to identify a proper composition range that meets all property requirements. In this work, we developed a CALPHAD-based ICME framework (CALPHAD: calculations of phase diagrams, ICME: integrated computational materials engineering) to optimize the composition and perform uncertainty quantifications, using high-strength low-alloy (HSLA) steel as a case study. Critical properties, such as impact transition temperature, yield strength, and printability were evaluated. With the same uncertainty as initial composition, a new nominal composition was determined, and it increased the probability of successful builds by 47%. Feedstock produced based on the designed composition was printed and systematically characterized. The printed samples exhibited high room temperature strength, good printability, and excellent toughness at -20°C.