just to clarify we didn't align from sc1 to sc2. to be frank we are too lazy to do that.   but i found when i use the same algorithm to first partition sc1 and sc2 networks, for 4_coexpr_anonym_aligned_v2.txt i found **20% less modules**. our method depends on random seed, so i rerun it for several times, and it is **consistently 20% less**. for 1_ppi_anonym_aligned_v2.txt it is **10% less **modules. for 5_cancer_anonym_aligned_v2.txt it is **20% less**. and 6_homology **10% less**. the other two are the same. but they are much smaller and maybe less likely to activate a bug.   i don't think it is coincidence. i think there is a subset of genes mis-aligned in sc2, and thus caused low performance in sc2.   i say subset because i hand checked a couple, they are correct, but it doesn't mean they are all correct. and as i said, i am too lazy to really align the two.   happy to provide organizers this piece of code . but essentially i just changed one line that reads in different files.   or the networks are not sorted randomly? because i found my algorithm can be different depending on whether i read in the larger values in first.