You can follow Rod's method at fairly low pressures e.g. up to 10bar which is the output from a normal gas regulator, you don't need to go very high, it's really enough that it can come back out when you open the valve.

If you want a degree of accuracy you really need to keep as many things as possible the same between your samples and calibration standards, but ultimately the detector detects whatever reaches it. It's easy enough to calculate the weight of analyte in e.g a 1ul liquid injection at a conc of say 100ppm then work this back from whatever response you get from your gas samples to give your gas composition.