I am very new to phylogenetic tree construction and am looking for advice on my method for choosing appropriate outgroups. I'm looking at constructing phylogenetic trees for Mycobacterium Tuberculosis paralog families. I have been using a low stringency nucleotide blast to find orthologs to the genes in my gene families and using these as the outgroups. A guiding principle, as I understand it, for outgroup choice is that the outgroup sequence should be related to the members of the gene family but not more so than any of the family members are to each other. My problem is that I am unsure how to judge this criterion and always seem to obtain trees that contain the outgroup in a cluster with one of the members of the family when rooting on the midpoint. Should I simply ignore this and root on the branch with the outgroup I have chosen or is this an indication that the outgroup and the family sequence are too closely related?
Thank you for your patience.
EDIT: I'm using the ML construction method with protein coding nucleotide sequences using a multiple codon alignment (MUSCLE) in MEGA6.