Depending on how many variations there are, you might be better off creating a data mapping spreadsheet, that will list all variations of the original text against the 'desired value', then merge it with your dataset.
Demo and explanation here:
http://forums.visokio.com/discussion/2365

Omniscope has function
LEFT(text, num_chars)
that might be helpful to create a formula, but search/replace later might be inefficient, in case "PT" appears after the first 2 characters.

Hi Paola,

Thanks for the comments.

I used the following formula assuming PTetc doesn't come anywhere else in the text:

IF(CONTAINS([Pivot series],"PT"),REPLACE([Pivot series],0, 2, "m1"),
IF(CONTAINS([Pivot series],"m1"),REPLACE([Pivot series],0, 2, "m2"),
IF(CONTAINS([Pivot series],"m2"),REPLACE([Pivot series],0, 2, "m3"),
IF(CONTAINS([Pivot series],"m3"),REPLACE([Pivot series],0, 2, "m4"),
IF(CONTAINS([Pivot series],"m4"),REPLACE([Pivot series],0, 2, "m5"),
IF(CONTAINS([Pivot series],"m5"),REPLACE([Pivot series],0, 2, "m6"),
IF(CONTAINS([Pivot series],"m6"),REPLACE([Pivot series],0, 2, "m7"),"")))))))

However, I thought there might be a more elegant solution.

Phillipa

I would suggest using startswith() instead of contains.

Thanks Daniel. I wasn't aware of the startswith function.