Null hypothesis is the opposite of study hypothesis or the one which states no difference My study hypothesis is that drug A will be at least 10% better than drug B. The most appropriate null hypothesis for this should be 1 or 2?

*

*Drug A is not at least 10% better than drug B (i.e., the opposite of the study hypothesis)

*Drug A and drug B are not different (0% difference)

I believe that option 1 is correct. Am I right?
 A: The null hypothesis should be stated in a way that it
implies a specific probability distribution (called the 'null' distribution), which is used to compute the P-value.
In general, the alternative might be one-sided or two-sided.
Two-sided: $H_0: \mu_A = \mu_B$ against $H_a: \mu_A \ne \mu_B.$
Right-sided: $H_0: \mu_A \le \mu_B$ (or $H_0: \mu_A = \mu_B);$
against alternative $H_a: \mu_A > \mu_B.$
Left-sided: $H_0: \mu_A \ge \mu_B$ (or $H_0: \mu_A = \mu_B);$
against alternative $H_a: \mu_A < \mu_B.$
Based on the information provided, you need the left-sided alternative. Notice that the null hypothesis contains an $=$-sign
in all of the formulations---whether as $=, \ge,$ of $\le.$
[Different authors have various preferences about this.]
In any case, you should choose Answer (2).
The effect size (10% or whatever)
can be determined after you have data. If you want to be reasonably sure
to detect an effect of a particular size, then you need to do a 'power and sample size' procedure to see how many subjects should be used in the study.
I agree with @Pitouille that specifying the desired effect as "10%" better may lead to confusion. Ten percent of what? (More increase in red blood cells? decrease in blood sugar level? decrease in patient complaints of pain? increase in patient 'recovery"?
What will you do if the data from the study show a clear indication of "8%" better?
Abandon Drug A as useless? do a new study a double the dosage?
What will you do if Drug A shows slightly better results, but
not by quite enough to be statistically significant at the desired level?  These are important considerations that should influence the design of the study, but usually not for inclusion in the statement of $H_0.$
