DeepTMHMM: A Deep Learning Model for Transmembrane Topology Prediction and Classification.

The model encodes the primary amino acid sequence by a pre-trained language model and decodes the topology by a state space model to produce topology and type predictions at unprecedented accuracy. DeepTMHMM makes it possible to scan full proteomes in order to detect both classes of transmembrane proteins.

References:

• Jeppe Hallgren, Konstantinos D. Tsirigos, Mads D. Pedersen, José Juan Almagro Armenteros, Paolo Marcatili, Henrik Nielsen, Anders Krogh and Ole Winther. DeepTMHMM predicts alpha and beta transmembrane proteins using deep neural networks. bioRxiv 2022.

• DeepTMHMM Main Site
• DeepTMHMM server

• Module Name: deeptmhmm (see the modules page for more information)
• GPU-capable. See GPU allocation guide.
• Environment variables set
  • DEEPTMHMM_HOME
• Example files in $DEEPTMHMM_HOME/test

Allocate an interactive session and run the program.
Sample session (user input in bold):

[user@biowulf]$ sinteractive
salloc.exe: Pending job allocation 46116226
salloc.exe: job 46116226 queued and waiting for resources
salloc.exe: job 46116226 has been allocated resources
salloc.exe: Granted job allocation 46116226
salloc.exe: Waiting for resource configuration
salloc.exe: Nodes cn3144 are ready for job

[user@cn3144 ~]$ module load deeptmhmm

[user@cn3144 ~]$ DeepTMHMM --fasta $DEEPTMHMM_HOME/test/sample.fasta --output-dir out 
Running DeepTMHMM on 1 sequence...
Step 1/4 | Loading transformer model...

Step 2/4 | Generating embeddings for sequences...
Generating embeddings: 100%|█████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:10<00:00, 10.31s/seq]

Step 3/4 | Predicting topologies for sequences in batches of 1...
Topology prediction: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:07<00:00,  7.18s/seq]

Step 4/4 | Generating output...

[user@cn3144 ~]$ ls out
TMRs.gff3  deeptmhmm_results.md  embeddings  plot.png  predicted_topologies.3line  probabilities
[user@cn3144 ~]$ exit
salloc.exe: Relinquishing job allocation 46116226
[user@biowulf ~]$

Create a batch input file (e.g. tmhmm.sh). For example:

#!/bin/bash
set -e
module load deeptmhmm
DeepTMHMM --fasta $DEEPTMHMM_HOME/test/sample.fasta --output-dir out

Submit this job using the Slurm sbatch command.

sbatch --gpus 1 [--cpus-per-task=#] [--mem=#] tmhmm.sh

Create a swarmfile (e.g. tmhmm.swarm). For example:

DeepTMHMM --fasta sample1.fasta --output-dir sample1_out
DeepTMHMM --fasta sample2.fasta --output-dir sample2_out
DeepTMHMM --fasta sample3.fasta --output-dir sample3_out
DeepTMHMM --fasta sample4.fasta --output-dir sample4_out

Submit this job using the swarm command.

swarm -f tmhmm.swarm [-g #] [-t #] --module deeptmhmm