The dry lab plan that we had for the Thioesterase – FAS complex integration modelling and the engineering of the efflux pump required us to work with structures of certain proteins that were not characterized in literature. Hence the first step in both the projects was to predict structures for all required proteins whose crystal structures were not available. We used Alphafold 2.0 for this purpose.[1]

Alphafold 2.0

AlphaFold 2.0 is an end-to-end solution for predicting protein structures. It takes a protein sequence as input and produces a 3D protein structure model in PDB format as output. The process involves sequence pre-processing to create a multiple-sequence alignment (MSA) and then using a neural network to generate 3D models. These models are further refined using force field calculations (AMBER), and confidence scores are provided to assess their quality. AlphaFold produces a per-residue estimate of its confidence on a scale from 0 - 100. This confidence measure is called pLDDT and corresponds to the model’s predicted score on the lDDT-Cα metric.[2]

For each prediction, slurm batch scripts were written to use the associated FASTA sequence of the protein as input and these were run on IISER Pune’s PARAM Brahma supercomputer cluster. Each prediction gave multiple structures as output, ranked based on its confidence score. In both the projects, the structural prediction ranked 0 (with highest confidence score) was taken as the structure for the associated protein and all the later steps were performed using these structures.

Two protein structures were predicted as part of the integration modelling and the efflux pump engineering:

The predicted structures, after the refining and removal of uncharacterized domains are shown below:

Structural Descriptions


JcFatB is structurally characterized to have a hotdog-fold. The basic subunits of hotdog -fold protein are 5–6 antiparallel beta sheets that are wrapped around an α-helix. These monomers dimerise, forming a homodimer, through interactions between their beta sheets leading to a continuous beta sheet and two antiparallel alpha helices.

ABC2 Transporter

ABC2 transporters have a core structure consisting of two transmembrane domains (TMDs) and two Nucleotide binding domains. TMDs contain helices that form the channel for the transport and are involved in the recognition of substrate while NBDs are the sites for ATP binding and hydrolysis, which powers the pump’s functioning.