Little is known about acetylation signaling pathways in early branching organisms such as trypanosomatids in comparison to other unicellular eucaryotic species like yeast. In addition to important biological differences described in this deep branched organisms, serious bioinformatic limitations arise when facing the highly divergent sequences of the proteins involved, making it difficult to perform homology-based prediction of protein functions. The availability of several interactomic datasets, like those recently generated by Staneva et al for Trypanosoma brucei acetyl-lysine readers, writers and erasers and the public release of AlphaFold2 (AF2) and AF2-multimer could be envisaged as a shortcut to address this problem. These tertiary and quaternary structure predictions could be used as a tool with high predictive value when inferring protein function. In this work, we made use of public interactomic datasets to predict direct interactions using AF2-multimer and validated them by yeast two-hybrid (Y2H) assays. We focused on MRG domain-containing proteins of Trypanosoma cruzi and their interactions, typically found as structural part of histone acetyl transferase (HAT) and histone deacetylase (HDAC) complexes. Our results led us to identify TcTINTIN, a structurally conserved complex that is orthologous to the human and yeast TINTIN complexes, that cannot be identified by homology searches based only in the primary sequences. Our in silico and wet lab approach led us to determine a specific sequential assembly of this complex, involving the proteins TcMRGx (MORF4 Related Gene X), TcMRGBP (MRG Binding Protein) and TcBDF6 (Bromodomain Factor 6). We also found another MRG domain assembled in a trimeric complex formed between TcBDF5, TcBDF5BP (BDF5 Binding Protein) and TcBDF8. In this complex we describe a novel way in which an MRG domain participates in the interaction with two proteins binding two different surfaces, instead of just one as previously reported. Together, these results show that AF2-processed interactomic datasets can be used to identify chromatin-remodeling protein complexes previously unknown in deeply branched eukaryotes.