Postdoctoral Researcher — University of Manitoba

A S M Zisanur Rahman

Computational Biologist & Bioinformatician

I build analysis pipelines and machine-learning models that turn sequencing and high-throughput screening data into biological insight: from metagenome-assembled genomes and resistome profiling in livestock systems to CRISPRi-seq chemical genomics and data-driven antibiotic discovery. Currently a Mitacs Accelerate Postdoctoral Fellow working with Topigs Norsvin on machine learning for feed efficiency and, through an NSERC Alliance project with Prof. Le Luo Guan (UBC), on machine learning and genome-scale metabolic models (GEMs) of individual rumen methanogens and the rumen community to reduce enteric methane.

Metagenomics Machine Learning CRISPRi-seq Antimicrobial Resistance Genome-Scale Metabolic Models HPC Workflows

View Projects Publications Contact

Expertise

What I Do

Metagenomics & Genome Recovery

Shotgun and culture-enriched metagenomics on Illumina and Nanopore data: assembly, ensemble binning into MAGs, taxonomic profiling, and resistome and mobilome quantification for AMR surveillance in agricultural systems.

MEGAHIT · metaWRAP-style binning · CheckM · MetaPhlAn · RGI · AMRFinder · mobileOG

Machine Learning for Discovery

Graph neural networks and tree-based models trained on high-throughput antibacterial screens and chemical–genetic interaction profiles to prioritise compounds, predict mechanism of action, and classify genomes by host phenotype.

Chemprop D-MPNN · scikit-learn · XGBoost · PyTorch · RDKit

CRISPRi-seq & Chemical Genomics

Design and analysis of pooled CRISPRi knockdown libraries screened against thousands of compounds, producing interaction profiles that reveal essential-gene function and antibiotic targets in Burkholderia.

sgRNA design · depletion modelling · t-SNE / clustering · Tanimoto similarity

Reproducible Pipelines on HPC

Modular, SLURM-ready workflows and interactive data apps that let collaborators re-run analyses and explore large datasets without touching the command line.

Python · R · Bash · Snakemake · SLURM · Shiny · Dash · Git

Ongoing Research

Current Projects

Two funded collaborations anchor my postdoctoral work: an industry partnership on feed efficiency and an NSERC Alliance project on enteric methane. Both apply machine learning to genome-resolved livestock microbiome data.

Industry partner Topigs Norsvin Mitacs Accelerate

AI-driven discovery of microbial functional signatures for feed efficiency in pigs

Host genetics explains only about a fifth of the variation in feed efficiency. We are building a functional reference database of the swine gut microbiome from more than 400 cultured isolates and 3,700 metagenome-assembled genomes, then training interpretable machine-learning models on functional gene profiles from Topigs Norsvin lines with divergent feed-efficiency phenotypes.

Metagenomic sampling of high- and low-feed-efficiency animals
Classical ML baselines plus tabular deep learning (FT-Transformer, TabNet)
SHAP and attention-based interpretation; leave-one-farm-out and cross-cohort validation

Snakemake · scikit-learn · XGBoost · PyTorch · SHAP · Optuna · Narval / Grex HPC

NSERC Alliance with Prof. Le Luo Guan, UBC

Machine learning and genome-scale metabolic models (GEMs) to reduce enteric methane from cattle

Enteric methane from ruminants is a major agricultural greenhouse-gas source, and Methanobrevibacter methanogens produce most of it. I lead the computational side of a collaboration with the Guan lab at the University of British Columbia that links rumen microbiome data and machine learning to methane phenotypes and builds genome-scale metabolic models (GEMs) at two levels: individual methanogens, to pinpoint essential reactions, and the rumen microbial community, to simulate hydrogen flow and methane output across species.

Individual GEMs for a ten-species rumen pan-methanogenome
Community-level rumen GEMs coupling methanogens with fermenting bacteria
Essential archaea-specific reactions prioritised as inhibitor targets
Nanopore rumen metagenomes and methanogen–bacteria co-culture assays

COBRApy · CarveMe · ModelSEED · RAVEN · memote · Nanopore · R

Computational drug discovery Anti-methanogen pipeline

Virtual screening of natural products against archaea-specific methanogen enzymes

Building on the methanogen GEMs, I developed an open, HPC-scale pipeline that docks a non-redundant library of ~792,000 natural products against 20 archaea-specific targets across methanogenesis and coenzyme F420, F430, CoM and H₄MPT biosynthesis, with molecular-dynamics follow-up of top hits and selectivity screened against the bovine proteome.

14 million docking runs orchestrated as SLURM array jobs
Automated ligand and receptor preparation with RDKit, Meeko and PDBFixer
F420-autofluorescence assay for validating anti-methanogenic hits

AutoDock Vina · RDKit · OpenMM · AlphaFold structures · SLURM

Software & Pipelines

Featured Projects

MAGnet

Shell · SLURM

Modular, HPC-friendly metagenomic binning and refinement suite that recovers high-quality MAGs with ensemble binning and consensus refinement. Runs on single assemblies or grouped co-assemblies, for Illumina and Nanopore reads.

MetagenomicsMAG recoveryNanopore

CIMPLE-Seq CGIP Explorer

R · Shiny

Interactive explorer for chemical–genetic interaction profiles of B. cenocepacia essential-gene knockdowns screened against ~5,000 compounds. Companion to my first-author Cell Reports paper.

CRISPRi-seqChemical genomicsInteractive app

Antibacterial Activity Prediction

Python · Chemprop

Directed message-passing neural network trained on a high-throughput antibacterial screen, with released weights and benchmarks. Published in PLOS Computational Biology, where it increased the hit rate of drug discovery.

Deep learningDrug discoveryCheminformatics

All projects → GitHub profile

Latest Work

Selected Publications

2026

Culture-enriched metagenomics recovers novel, high-quality genome catalogue and a host-linked virome of the cultivable bovine rumen microbiome

ASMZ Rahman, L Scott, A Alizadeh, RJ Gruninger, TA McAllister, LL Guan, H Derakhshani — Animal Microbiome

First authorSubmitted

2026

Genome-resolved profiling of an expanded swine gut isolate collection reveals functional signatures of health and disease

ASMZ Rahman, F Mohammadian Loojaly, A Farzan, MG Surette, S Walkowiak, C Yang, H Derakhshani — Microbiome

First authorAccepted, in press

2026

Optimizing a culture-enriched hybrid metagenomics pipeline to assess the AMR footprint of livestock manure in anaerobic digestate

N Rahman, ASMZ Rahman, DB Levin, TA McAllister, N Cicek, H Derakhshani — Microbiology Spectrum

Co-first author

2024

Rationally designed pooled CRISPRi-seq uncovers an inhibitor of bacterial peptidyl-tRNA hydrolase

ASMZ Rahman et al. — Cell Reports 43(11)

First authorCell Reports

2023

Profiling cell envelope-antibiotic interactions reveals vulnerabilities to β-lactams in a multidrug-resistant bacterium

AM Hogan, ASMZ Rahman, A Motnenko et al. — Nature Communications 14(1)

Nature Comms

2022

A machine learning model trained on a high-throughput antibacterial screen increases the hit rate of drug discovery

ASMZ Rahman, C Liu, H Sturm, AM Hogan et al. — PLOS Computational Biology 18(10)

First authorPLOS Comp Bio

All publications →