Hyderabad researcher develops AI tool to predict protein failure
HYDERABAD: A 24 year old scientist from Hyderabad has achieved a breakthrough in solving a long standing global biological question on how amino acid shortages disrupt protein production in the human body.
Using artificial intelligence, Mohan Vamsi Nallapareddy and his collaborators have developed a model that can forecast protein-production failures at the molecular level. Researchers say such predictive ability could significantly advance cancer treatment and genetic therapies.
Model identifies amino acid shortages in gene sequences
Every human cell relies on amino acids derived from food to produce essential proteins. In conditions such as malnutrition, chronic illness or fast-growing cancers, specific amino acids become deficient. When this happens, the cell’s protein-production machinery slows or stops, leading to incomplete or incorrect proteins. In severe cases, vital proteins fail to form, weakening immunity and worsening disease.
For decades, scientists struggled to pinpoint the exact genetic sequences responsible for such breakdowns. The new deep-learning model identifies these mechanisms with precision.
According to researchers, the tool can read the genetic code associated with any protein making gene and determine which amino acids are missing. It predicts exactly where protein synthesis will stall and recognises codon patterns linked to the disruption. This allows scientists to assess a gene sequence and detect potential failures before they occur in the body.
Implications for cancer biology and therapeutics
The model provides deeper insights into how nutritional deprivation shapes disease progression, particularly in cancer biology and metabolic disorders. It also supports the design of targeted gene therapies and artificial proteins. In the biopharmaceutical sector, it could improve manufacturing of insulin, vaccines, antibodies and other biologics.
Researchers say the findings also help explain why certain genetic mutations cause disease only under specific conditions.
The work was published in Communications Biology, a journal of the Nature group. “From here, I will focus on precision medicine and gene therapy,” Vamsi said.
Hyderabad educated scientist now at EPFL
After schooling in Hyderabad and completing a BE in Computer Science from BITS Pilani, Vamsi worked as a research assistant at University College London. He is now a doctoral assistant at the Ecole Polytechnique Federale de Lausanne in Switzerland, guided by Pierre Vandergheynst, an expert in convolutional neural networks on graphs.
