Regex to remove unwanted text in gene sequences - regex

Question

I have gene sequences that can have actual string text in them I want to remove with regex. I would like to try to remove the errant text in a generic way with regex. I'd like to remove all characters up to 10 chars between any invalid characters. I am assuming that anything between invalid chars up to 10 chars apart is part of the invalid text.
example :
BADTEXTATTHEBEGINNINGATCATCGGCCCATGCATMOREBADTEXTINTHEMIDDLEGCGGGGATCGCCCCTTTAAAATHISISSOMETEXTATTHEENDIWANTREMOVED
Valid sequence characters are ATCG. Can we create a regex to reduce the above string to
GATCATCGGCCCATGCATGCGGGGATCGCCCCTTTAAAAT?
I understand that the G at the beginning of this final sequence is the last character of the word BEGINNING, which is the "bad" text at the beginning of the string. I realize with regex, it is impossible to identify words, so I am willing to live this limitation. Same with the T at the end, which is the first letter of "THIS".
I've tried to do something with repeated capture groups that allow for a certain number of chars between bad characters, but I can't seem to make it work right. Maybe someone can help me...
This regex does not quite work to capture everything.
([^ACTG].{1,10}[^ACTG])+

Answer 1

Initial string:
BADTEXTATTHEBEGINNINGATCATCGGCCCATGCATMOREBADTEXTINTHEMIDDLEGCGGGGATCGCCCCTTTAAAATHISISSOMETEXTATTHEENDIWANTREMOVED
String after replacing non-ACGT:
-A-T--TATT----G-----GATCATCGGCCCATGCAT-----A-T--T--T--------GCGGGGATCGCCCCTTTAAAAT---------T--TATT-------A-T-------
For this sample, a run of up to four ACGT characters can appear in the unwanted text. Examining other samples may give a sensible upper bound.
Perhaps "starts and ends with invalid character and contains no long runs of valid characters" is a better measure to use than "1 to 10 characters, starting and ending with invalid character"?
A regex for this is:
[^ACGT]((?![ACGT]{5,}).)*[^ACGT]
and matches:
BADTEXTATTHEBEGINNIN
MOREBADTEXTINTHEMIDDLE
HISISSOMETEXTATTHEENDIWANTREMOVED