search

baker-laboratory / RoseTTAFold-All-Atom

Other
653 stars 118 forks source link

Minor bug affecting expanding multi-MSAs #47

Closed milafternoon closed 8 months ago

milafternoon commented 8 months ago

Hi,

Thanks a lot for releasing the code! This is just to report a tiny bug that prevented the correct construction of an expanded MSA for a mixed protein homo/heterotrimer (a homodimer bound to a monomer).

If I'm not mistaken, line 48 in data/merge_inputs.py should read: a3m_out = join_msas_by_taxid(a3m_out, unique_a3m[i])

amorehead commented 8 months ago

That does seem to be the case, at least from my perspective as well.

jayunruh commented 8 months ago

In a related issue it also appears that line 52 in data/merge_inputs.py should read a3m_out['a3m'],a3m_out['ins'] = merge_a3m_homo(msa, ins, len(hash_list)) or alternatively merge_a3m_homo should be changed to output a dictionary.

milafternoon commented 8 months ago

Confirmed, I made the same change myself (first defining a3m_out as a copy of a3m), that was already pointed out here though: https://github.com/baker-laboratory/RoseTTAFold-All-Atom/issues/35