Performance in OpenFold3 NIM

NIM Accuracy

OpenFold3 is an all-atom biomolecular complex structure prediction model from the OpenFold Consortium and the AlQuraishi Laboratory. OpenFold3 is a PyTorch implementation of the jax-based AlphaFold3 reported in Accurate structure prediction of biomolecular interactions with AlphaFold 3, and like AlphaFold3, OpenFold3 extends protein structure prediction capabilities to model complete biomolecular complexes including proteins, DNA, RNA, and small molecule ligands.

The OpenFold3 NIM’s accuracy should match that of the AlQuraishi Laboratory implementation of OpenFold3, when using equivalent parameters and inputs.

Note

Running on hardware that is not listed as supported in the prerequisites section may produce results that deviate from the expected accuracy.

The accuracy of the NIM is measured by structural quality metrics such as lddt (local distance difference test). These scores help assess the reliability of the predicted structures.

Factors Affecting NIM Performance

The performance of the OpenFold3 NIM is determined by several key factors:

Hardware Factors

• GPU type and memory: Different GPU architectures provide different performance levels

• System RAM: Larger proteins and complexes require more system memory

• Storage speed: Fast NVMe SSD storage improves model loading and caching performance

Input Complexity

• Sequence length: Runtime increases with total sequence length

• Number of chains: Multi-chain complexes require more computation than single chains

• MSA size: Larger MSAs can improve accuracy but increase memory usage and computation time

• Ligands, DNA, RNA: Additional molecular components increase computational cost

Model Configuration

• Diffusion samples: Multiple samples provide diversity but multiply computational cost

Performance Characteristics

Typical Runtimes

For reference, approximate runtimes on high-end hardware (NVIDIA H100 80GB):

Structure Prediction (OpenFold3 NIM v1.6.0 on NVIDIA H100 80GB):

• ~200 residues: 2.9-3.6 seconds

• ~300-400 residues: 4.5-12.6 seconds

• ~500-600 residues: 11.6-26.8 seconds

• ~800-900 residues: 20.6 seconds

• ~1300-1500 residues: 30.5-53.1 seconds

• ~1700-1900 residues: 56.5-72.5 seconds

• Memory usage: Varies with sequence length, typically 40-80GB GPU memory

Note

Short sequences improved substantially in 1.6.0. In earlier releases, runtime below roughly 400 residues was dominated by fixed per-request overhead rather than by the model itself, so a 186-residue input cost almost as much as a 600-residue one. CUDA graph capture removes most of that overhead, so runtime now tracks sequence length across the whole range.

Performance Notes The total runtime for structure prediction depends on:

• Total number of residues in the complex

• Total number of atoms in the complex

• Number of molecules and chains

• Number of sequences in the MSAs

• Number of diffusion samples requested

Performance Results on H100

The following table shows OpenFold3 NIM runtime on NVIDIA H100 80GB, with and without structural template processing.

Note

Structural template support is available starting from version 1.1.0.

Test ID	Seq Length	Without structural templates (s)	With structural templates (s)
8eil	186	2.93	4.62
7r6r	203	3.64	4.94
1a3n	287	12.63	22.54
8c4d	331	4.55	5.01
7qsj	375	5.80	6.57
8cpk	384	9.21	10.98
8are	530	11.56	12.47
8owf	575	12.68	12.94
8aw3	590	26.84	28.34
7tpu	616	9.76	10.66
7ylz	623	17.54	18.80
8gpp	628	15.28	21.33
8clz	684	16.36	17.96
8k7x	858	20.62	22.30
8ibx	1286	30.46	32.26
8gi1	1464	43.93	46.70
8sm6	1496	53.12	58.54
8pso	1499	42.72	45.76
8jue	1657	56.52	61.74
8bsh	1762	63.19	68.27
5xgo	1869	72.52	87.65

All runtimes are in seconds for end-to-end structure prediction with a single diffusion sample. The template measurements include an average of 4 CIF template files per protein chain. The additional time is primarily attributed to CIF parsing.

Performance Analysis

Key Observations:

• Scaling Behavior: Runtime increases with total sequence length and with the number of chains and atoms in the complex.

• Change from 1.5.0: CUDA graph capture removed a fixed per-request cost of roughly 7-8 seconds on H100, so comparing these numbers with the 1.5.0 table shows a ratio that shrinks as inputs grow, from about 3.5x at 186 residues to about 1.15x at 1869, even though the same amount of time was saved in both cases. Read the difference between the two releases, not the ratio.

• Hardware range: The fastest SKUs (GB200, GB300) complete typical predictions in roughly 75% of the H100 time, with B200, H200 and GH200 at 88-91%. A100-class parts take 1.4-1.7x the H100 time. Refer to Table 1 for per-SKU numbers.

• Small proteins (<400 residues): Complete in ~3-13 seconds on H100.

• Large proteins (>1500 residues): Complete in ~43-73 seconds on H100.

Template Processing Impact:

• Template overhead: Structural templates add modest overhead (typically 1-9 seconds) depending on sequence length and chain count.

• Primary cost: The additional time is mostly attributed to CIF parsing (average 4 templates per chain).

• Overall impact: Templates provide structural guidance with minimal performance cost.

Recommended Configuration:

• For large proteins (>1800 residues): Prefer high-memory GPUs; actual requirements depend on workload (MSA depth, templates, and molecule composition). Refer to Table 1 for measured per-SKU results.

• For template-guided predictions: Structural templates add minimal performance overhead.

Configuration

The benchmarks use the following configuration:

Parameter	Setting
diffusion_samples	1
output_format	pdb
GPU	H100 80GB
structural_templates	Average 4 CIF files per chain

Performance Metrics

The following table contains end-to-end runtime (seconds) for the OpenFold3 NIM across supported NVIDIA hardware units. Inputs are the same 21 benchmark cases, arranged by sequence length and annotated with PDB ID and sequence length.

Table 1: Performance Across the Supported NVIDIA Hardware Units

The table shows the prediction times for tests with varying sequence lengths for hardware units. It features the following:

• Hardware: Nineteen supported GPU SKUs (NVIDIA H100 80GB HBM3, H100 NVL, H100 PCIe, H200, H200 NVL, B200, B300, GB200, GB300, GH200, A100 SXM4 80GB, A100 80GB PCIe, A100 SXM4 40GB, A100 40GB PCIe, L40S, RTX PRO 6000 Blackwell Server Edition, RTX PRO 6000 Blackwell Workstation Edition, RTX 6000 Ada Generation, and DGX Spark GB10). One tab per SKU.

• Metric: End-to-end predict time (seconds).

• Configuration: Default model; no structural templates. Inputs sorted by sequence length and annotated with PDB ID and sequence length.

Note

A dash (—) means the case was not measured on that SKU. The benchmark suite caps input length on memory-constrained parts: 1500 residues on the 40-48 GB GPUs (L40S, RTX 6000 Ada, A100 SXM4 40GB, A100 40GB PCIe) and 1800 residues on GB10. These are benchmark limits, not limits the NIM enforces. The NIM itself imposes no maximum sequence length. Longer inputs on those GPUs may still run, but are not covered by these measurements.

NVIDIA H100 80GB HBM3

Test ID	Seq Length	predict_time (s)
8eil	186	2.93
7r6r	203	3.64
1a3n	287	12.63
8c4d	331	4.55
7qsj	375	5.80
8cpk	384	9.21
8are	530	11.56
8owf	575	12.68
8aw3	590	26.84
7tpu	616	9.76
7ylz	623	17.54
8gpp	628	15.28
8clz	684	16.36
8k7x	858	20.62
8ibx	1286	30.46
8gi1	1464	43.93
8sm6	1496	53.12
8pso	1499	42.72
8jue	1657	56.52
8bsh	1762	63.19
5xgo	1869	72.52

NVIDIA H100 NVL

Test ID	Seq Length	predict_time (s)
8eil	186	2.86
7r6r	203	3.56
1a3n	287	12.37
8c4d	331	4.45
7qsj	375	5.73
8cpk	384	8.74
8are	530	11.24
8owf	575	12.51
8aw3	590	25.93
7tpu	616	9.69
7ylz	623	16.92
8gpp	628	14.81
8clz	684	15.85
8k7x	858	20.94
8ibx	1286	30.33
8gi1	1464	46.15
8sm6	1496	55.16
8pso	1499	44.96
8jue	1657	59.89
8bsh	1762	68.00
5xgo	1869	78.09

NVIDIA H100 PCIe

Test ID	Seq Length	predict_time (s)
8eil	186	3.06
7r6r	203	3.96
1a3n	287	14.23
8c4d	331	5.12
7qsj	375	6.58
8cpk	384	10.44
8are	530	13.07
8owf	575	14.18
8aw3	590	29.84
7tpu	616	11.69
7ylz	623	19.59
8gpp	628	17.08
8clz	684	18.50
8k7x	858	25.11
8ibx	1286	37.68
8gi1	1464	56.81
8sm6	1496	66.62
8pso	1499	55.48
8jue	1657	73.93
8bsh	1762	82.75
5xgo	1869	95.95

NVIDIA H200

Test ID	Seq Length	predict_time (s)
8eil	186	2.63
7r6r	203	3.26
1a3n	287	11.34
8c4d	331	4.09
7qsj	375	5.27
8cpk	384	8.08
8are	530	10.22
8owf	575	11.38
8aw3	590	23.46
7tpu	616	8.72
7ylz	623	15.46
8gpp	628	13.48
8clz	684	14.61
8k7x	858	18.60
8ibx	1286	26.43
8gi1	1464	39.14
8sm6	1496	47.36
8pso	1499	38.68
8jue	1657	51.20
8bsh	1762	57.24
5xgo	1869	66.04

NVIDIA H200 NVL

Test ID	Seq Length	predict_time (s)
8eil	186	2.70
7r6r	203	3.41
1a3n	287	11.49
8c4d	331	4.34
7qsj	375	5.47
8cpk	384	8.51
8are	530	10.52
8owf	575	11.69
8aw3	590	23.78
7tpu	616	9.08
7ylz	623	15.59
8gpp	628	13.98
8clz	684	15.09
8k7x	858	19.44
8ibx	1286	28.37
8gi1	1464	42.25
8sm6	1496	50.46
8pso	1499	42.09
8jue	1657	55.16
8bsh	1762	61.68
5xgo	1869	71.70

NVIDIA B200

Test ID	Seq Length	predict_time (s)
8eil	186	2.62
7r6r	203	3.16
1a3n	287	11.27
8c4d	331	3.98
7qsj	375	4.95
8cpk	384	7.87
8are	530	10.35
8owf	575	10.99
8aw3	590	23.56
7tpu	616	8.15
7ylz	623	15.75
8gpp	628	13.60
8clz	684	14.54
8k7x	858	19.37
8ibx	1286	28.07
8gi1	1464	34.75
8sm6	1496	43.01
8pso	1499	41.55
8jue	1657	56.35
8bsh	1762	63.47
5xgo	1869	74.35

NVIDIA B300 SXM6

Test ID	Seq Length	predict_time (s)
8eil	186	2.49
7r6r	203	2.95
1a3n	287	11.41
8c4d	331	3.86
7qsj	375	5.03
8cpk	384	7.92
8are	530	13.65
8owf	575	14.42
8aw3	590	23.09
7tpu	616	8.27
7ylz	623	15.53
8gpp	628	17.61
8clz	684	18.46
8k7x	858	23.19
8ibx	1286	32.76
8gi1	1464	39.51
8sm6	1496	51.06
8pso	1499	40.97
8jue	1657	55.87
8bsh	1762	63.18
5xgo	1869	73.23

NVIDIA GB200

Test ID	Seq Length	predict_time (s)
8eil	186	2.73
7r6r	203	3.23
1a3n	287	9.47
8c4d	331	3.71
7qsj	375	4.57
8cpk	384	6.49
8are	530	8.48
8owf	575	9.43
8aw3	590	19.40
7tpu	616	7.08
7ylz	623	12.29
8gpp	628	10.75
8clz	684	11.56
8k7x	858	15.32
8ibx	1286	22.52
8gi1	1464	27.58
8sm6	1496	33.20
8pso	1499	34.19
8jue	1657	46.35
8bsh	1762	52.04
5xgo	1869	61.60

NVIDIA GB300

Test ID	Seq Length	predict_time (s)
8eil	186	2.82
7r6r	203	3.25
1a3n	287	9.51
8c4d	331	3.79
7qsj	375	4.62
8cpk	384	6.63
8are	530	8.56
8owf	575	9.51
8aw3	590	19.43
7tpu	616	7.09
7ylz	623	12.33
8gpp	628	10.96
8clz	684	11.65
8k7x	858	15.40
8ibx	1286	22.58
8gi1	1464	27.67
8sm6	1496	33.17
8pso	1499	34.12
8jue	1657	46.28
8bsh	1762	52.15
5xgo	1869	61.44

NVIDIA GH200 144GB HBM3e

Test ID	Seq Length	predict_time (s)
8eil	186	2.85
7r6r	203	3.31
1a3n	287	10.55
8c4d	331	4.24
7qsj	375	5.27
8cpk	384	7.03
8are	530	9.76
8owf	575	10.71
8aw3	590	22.62
7tpu	616	9.54
7ylz	623	14.14
8gpp	628	12.71
8clz	684	13.98
8k7x	858	19.01
8ibx	1286	29.27
8gi1	1464	47.74
8sm6	1496	53.82
8pso	1499	46.22
8jue	1657	62.32
8bsh	1762	71.89
5xgo	1869	82.37

NVIDIA A100 SXM4 80GB

Test ID	Seq Length	predict_time (s)
8eil	186	4.22
7r6r	203	5.02
1a3n	287	16.79
8c4d	331	6.60
7qsj	375	8.24
8cpk	384	12.10
8are	530	15.61
8owf	575	17.03
8aw3	590	34.98
7tpu	616	14.49
7ylz	623	22.80
8gpp	628	20.59
8clz	684	22.35
8k7x	858	29.62
8ibx	1286	43.61
8gi1	1464	66.33
8sm6	1496	77.76
8pso	1499	64.84
8jue	1657	87.04
8bsh	1762	95.62
5xgo	1869	114.93

NVIDIA A100 80GB PCIe

Test ID	Seq Length	predict_time (s)
8eil	186	3.61
7r6r	203	4.73
1a3n	287	16.57
8c4d	331	6.25
7qsj	375	7.97
8cpk	384	12.17
8are	530	15.13
8owf	575	16.62
8aw3	590	34.70
7tpu	616	14.22
7ylz	623	23.24
8gpp	628	20.64
8clz	684	22.47
8k7x	858	30.17
8ibx	1286	44.84
8gi1	1464	69.88
8sm6	1496	82.40
8pso	1499	68.07
8jue	1657	91.98
8bsh	1762	101.94
5xgo	1869	121.19

NVIDIA A100 SXM4 40GB

Test ID	Seq Length	predict_time (s)
8eil	186	4.38
7r6r	203	5.62
1a3n	287	19.67
8c4d	331	7.27
7qsj	375	9.43
8cpk	384	14.38
8are	530	17.56
8owf	575	19.03
8aw3	590	41.09
7tpu	616	16.06
7ylz	623	26.55
8gpp	628	23.43
8clz	684	25.23
8k7x	858	33.33
8ibx	1286	48.26
8gi1	1464	73.28
8sm6	1496	87.23
8pso	1499	71.20
8jue	1657	—
8bsh	1762	—
5xgo	1869	—

NVIDIA A100 40GB PCIe

Test ID	Seq Length	predict_time (s)
8eil	186	4.23
7r6r	203	5.53
1a3n	287	20.30
8c4d	331	7.33
7qsj	375	9.47
8cpk	384	14.60
8are	530	18.56
8owf	575	20.08
8aw3	590	43.77
7tpu	616	16.89
7ylz	623	28.38
8gpp	628	25.14
8clz	684	26.98
8k7x	858	35.76
8ibx	1286	53.41
8gi1	1464	80.47
8sm6	1496	95.66
8pso	1499	78.19
8jue	1657	—
8bsh	1762	—
5xgo	1869	—

NVIDIA L40S

Test ID	Seq Length	predict_time (s)
8eil	186	2.98
7r6r	203	3.92
1a3n	287	19.15
8c4d	331	5.81
7qsj	375	7.87
8cpk	384	11.74
8are	530	17.06
8owf	575	19.21
8aw3	590	40.26
7tpu	616	17.08
7ylz	623	25.96
8gpp	628	23.47
8clz	684	25.78
8k7x	858	35.98
8ibx	1286	58.12
8gi1	1464	93.17
8sm6	1496	106.68
8pso	1499	87.18
8jue	1657	—
8bsh	1762	—
5xgo	1869	—

NVIDIA RTX PRO 6000 Blackwell Server Edition

Test ID	Seq Length	predict_time (s)
8eil	186	2.79
7r6r	203	3.37
1a3n	287	13.78
8c4d	331	4.50
7qsj	375	5.76
8cpk	384	9.17
8are	530	12.36
8owf	575	13.97
8aw3	590	28.57
7tpu	616	11.38
7ylz	623	18.88
8gpp	628	16.59
8clz	684	18.06
8k7x	858	24.48
8ibx	1286	37.60
8gi1	1464	57.13
8sm6	1496	67.16
8pso	1499	57.96
8jue	1657	78.68
8bsh	1762	86.09
5xgo	1869	102.24

NVIDIA RTX PRO 6000 Blackwell Workstation Edition

Test ID	Seq Length	predict_time (s)
8eil	186	2.33
7r6r	203	2.86
1a3n	287	12.21
8c4d	331	3.88
7qsj	375	5.01
8cpk	384	8.34
8are	530	10.92
8owf	575	11.98
8aw3	590	25.72
7tpu	616	10.61
7ylz	623	17.22
8gpp	628	14.94
8clz	684	16.43
8k7x	858	22.64
8ibx	1286	35.02
8gi1	1464	55.24
8sm6	1496	64.27
8pso	1499	55.12
8jue	1657	75.62
8bsh	1762	83.73
5xgo	1869	98.91

NVIDIA RTX 6000 Ada Generation

Test ID	Seq Length	predict_time (s)
8eil	186	2.90
7r6r	203	3.73
1a3n	287	18.10
8c4d	331	5.59
7qsj	375	7.66
8cpk	384	11.88
8are	530	16.46
8owf	575	18.02
8aw3	590	38.80
7tpu	616	16.17
7ylz	623	25.11
8gpp	628	22.46
8clz	684	25.05
8k7x	858	35.61
8ibx	1286	54.70
8gi1	1464	88.55
8sm6	1496	101.91
8pso	1499	84.60
8jue	1657	—
8bsh	1762	—
5xgo	1869	—

NVIDIA DGX Spark (GB10)

Test ID	Seq Length	predict_time (s)
8eil	186	7.01
7r6r	203	11.69
1a3n	287	51.76
8c4d	331	18.41
7qsj	375	23.53
8cpk	384	25.21
8are	530	45.11
8owf	575	53.42
8aw3	590	94.69
7tpu	616	56.78
7ylz	623	65.16
8gpp	628	62.94
8clz	684	72.86
8k7x	858	111.42
8ibx	1286	184.24
8gi1	1464	456.44
8sm6	1496	511.18
8pso	1499	421.23
8jue	1657	657.04
8bsh	1762	751.01
5xgo	1869	—

Performance Optimization Tips

• GPU Selection: Use any of the supported GPUs listed in Table 1. For best throughput, prefer higher-performance SKUs such as B200, H100, and H200.

• Sequence Length: Performance scales with sequence length.

• Multiple Samples: Setting diffusion_samples > 1 increases runtime in affine fashion (input featurization time is independent of diffusion_samples).

• MSA Size: While larger MSAs can improve accuracy, they also increase memory usage and computation time. Consider filtering MSAs for very large proteins.

• Structural Templates: Templates add modest overhead (1-10 seconds) but can significantly improve prediction accuracy. Refer to Template Processing for guidance.

• Batch Processing: For multiple independent predictions, process them sequentially or use multiple NIM instances.

• Memory Management: Ensure adequate GPU memory for your target sequence lengths. Very long sequences (>1800 residues) are memory-intensive, and actual requirements depend on workload characteristics (MSA depth, templates, and molecule composition). Refer to Table 1 for measured per-SKU results.

Reproducing Performance Benchmarks

Overview

This section provides scripts and instructions for reproducing the performance metrics reported above. The benchmarking process is as follows:

1. Run inference with OpenFold3 NIM to generate predicted structures

2. Use OpenStructure (OST) to compare predictions against reference structures

3. Extract accuracy metrics like lDDT (local Distance Difference Test)

Prerequisites

Run the OpenFold3 NIM

Before running benchmarks, ensure that OpenFold3 NIM is deployed and running. The benchmarking scripts send inference requests to the NIM service.

Verify NIM is Running:

# Check if NIM is accessible and ready
curl http://localhost:8000/v1/health/ready

# Expected response:
# {"object":"health.response","message":"ready","status":"ready"}

Note

The default NIM URL is http://localhost:8000. If your NIM is running on a different host or port, you’ll need to specify it using the --nim-url parameter when running benchmarks. For detailed deployment instructions and configuration options, refer to the Getting Started guide.

Install the OpenStructure Docker Image

OpenStructure is a computational structural biology framework that provides tools for structure comparison and validation. You’ll need the OpenStructure Docker image for running benchmarks:

# Pull the latest OpenStructure image from the OST registry
docker pull registry.scicore.unibas.ch/schwede/openstructure:latest

# Verify the installation
docker run --rm -v $(pwd):/home registry.scicore.unibas.ch/schwede/openstructure:latest --version

# Expected response:
# OpenStructure 2.11.1

Download Reference Structures

You’ll need reference structures (ground truth) in CIF format for comparison. These are experimentally determined structures from the Protein Data Bank (PDB):

• CIF (Crystallographic Information File): A standard format for representing molecular structures, including atomic coordinates, experimental metadata, and structural annotations

• Obtaining reference structures: Download from RCSB PDB using the PDB IDs from the performance table (e.g., 8eil, 7r6r, 1a3n)

Example:

# reate directories
mkdir -p references

# Download a reference structure (e.g., 8eil) to references folder
wget https://files.rcsb.org/download/8eil.cif -O references/8eil.cif

Preparing Input JSON Files

What are Input JSON Files?

Input JSON files contain the sequence information needed for OpenFold3 NIM to make predictions. Each file specifies:

• Protein sequences (amino acid chains)

• Chain IDs

• Optional: DNA, RNA, ligands, MSAs, templates

Example Input JSON Structure

{
  "name": "8eil",
  "sequences": [
    {
      "protein": {
        "id": "A",
        "sequence": "MKQHKAMIVALIVICITAVVAALVTRKDLCEVHIRTGQTEVAVF..."
      }
    }
  ]
}

Create Input JSON Files

You can create input JSON files by extracting sequences from PDB structures. Save this script as generate_inputs.py:

#!/usr/bin/env python3
"""Generate input JSON files from PDB structures."""
import argparse
import json
import requests
import sys
from pathlib import Path

def fetch_sequences_from_pdb(pdb_id):
    """Fetch protein sequences from RCSB PDB FASTA endpoint."""
    url = f"https://www.rcsb.org/fasta/entry/{pdb_id}"
    response = requests.get(url)
    if response.status_code != 200:
        raise Exception(f"Failed to fetch FASTA for {pdb_id}")
    
    sequences = []
    current_seq = []
    chain_index = 0
    
    for line in response.text.strip().split('\n'):
        if line.startswith('>'):
            if current_seq:
                # Use sequential chain IDs: A, B, C, D, ...
                chain_id = chr(65 + chain_index)
                sequences.append({
                    "protein": {
                        "id": chain_id,
                        "sequence": ''.join(current_seq)
                    }
                })
                chain_index += 1
            current_seq = []
        else:
            current_seq.append(line.strip())
    
    # Don't forget the last sequence
    if current_seq:
        chain_id = chr(65 + chain_index)
        sequences.append({
            "protein": {
                "id": chain_id,
                "sequence": ''.join(current_seq)
            }
        })
    
    return sequences

def create_input_json(pdb_id):
    """Create OpenFold3 input JSON file for a PDB structure."""
    output_dir = Path("inputs")
    output_dir.mkdir(parents=True, exist_ok=True)
    
    print(f"Fetching sequences for {pdb_id}...")
    sequences = fetch_sequences_from_pdb(pdb_id)
    
    input_data = {
        "name": pdb_id,
        "sequences": sequences
    }
    
    output_path = output_dir / f"{pdb_id}_input.json"
    with open(output_path, 'w') as f:
        json.dump(input_data, f, indent=2)
    
    print(f"Created {output_path} ({len(sequences)} chain(s))")
    return output_path

if __name__ == "__main__":
    # All benchmark PDB IDs from performance table
    benchmark_pdbs = [
        "8eil", "7r6r", "1a3n", "8c4d", "7qsj", "8cpk", 
        "8are", "8owf", "8aw3", "7tpu", "7ylz", "8gpp", 
        "8clz", "8k7x", "8ibx", "8gi1", "8sm6", "8pso", 
        "8jue", "8bsh", "5xgo"
    ]
    
    parser = argparse.ArgumentParser(
        description='Generate OpenFold3 input JSON files from PDB structures',
        epilog='Example: python generate_inputs.py --pdb 8eil'
    )
    parser.add_argument('--pdb', type=str, 
                        help='Specific PDB ID to generate input for (e.g., 8eil)')
    parser.add_argument('--all', action='store_true',
                        help='Generate input files for all benchmark PDB IDs')
    
    args = parser.parse_args()
    
    # For backwards compatibility: if no arguments provided, generate all
    if not args.pdb and not args.all:
        args.all = True
    
    if args.pdb:
        # Generate input for specific PDB ID
        try:
            create_input_json(args.pdb)
        except Exception as e:
            print(f"Error processing {args.pdb}: {e}", file=sys.stderr)
            sys.exit(1)
    elif args.all:
        # Generate all benchmark inputs
        print("Generating input files for all benchmark cases...")
        failed = []
        for pdb_id in benchmark_pdbs:
            try:
                create_input_json(pdb_id)
            except Exception as e:
                print(f"Error processing {pdb_id}: {e}")
                failed.append(pdb_id)
        
        if failed:
            print(f"\nFailed to generate inputs for: {', '.join(failed)}", file=sys.stderr)
            sys.exit(1)

Benchmarking Script

The following is a complete script to benchmark OpenFold3 NIM predictions. Save this as benchmark_openfold3.py:

#!/usr/bin/env python3
"""
Benchmark OpenFold3 NIM predictions against reference structures.
"""

import argparse
import json
import subprocess
import time
from pathlib import Path
import requests

def run_inference(nim_url, input_json, output_dir):
    """
    Run OpenFold3 NIM inference and save the predicted structure.
    
    Args:
        nim_url: URL of the OpenFold3 NIM service (e.g., http://localhost:8000)
        input_json: Path to input JSON file with sequence information
        output_dir: Directory to save predicted structures
    
    Returns:
        tuple: (predicted_pdb_path, inference_time_seconds)
    """
    output_dir = Path(output_dir)
    output_dir.mkdir(parents=True, exist_ok=True)
    
    # Read input configuration
    with open(input_json, 'r') as f:
        input_data = json.load(f)
    
    # Convert to NIM API format
    molecules = []
    for seq in input_data.get("sequences", []):
        if "protein" in seq:
            protein = seq["protein"]
            # Create minimal MSA with just the query sequence
            msa_csv = f"key,sequence\n-1,{protein['sequence']}"
            molecules.append({
                "type": "protein",
                "id": [protein["id"]],
                "sequence": protein["sequence"],
                "msa": {
                    "main_db": {
                        "csv": {
                            "alignment": msa_csv,
                            "format": "csv"
                        }
                    }
                }
            })
    
    nim_request = {
        "inputs": [{
            "input_id": input_data.get("name", "prediction"),
            "molecules": molecules,
            "output_format": "pdb"
        }]
    }
    
    # Start timing
    start_time = time.time()
    
    # Run inference
    response = requests.post(
        f"{nim_url}/biology/openfold/openfold3/predict",
        json=nim_request,
        headers={"Content-Type": "application/json"}
    )
    
    # End timing
    inference_time = time.time() - start_time
    
    if response.status_code != 200:
        raise Exception(f"Inference failed: {response.text}")
    
    # Extract PDB data from response
    result = response.json()
    
    outputs = result.get('outputs', [])
    if not outputs:
        raise Exception(f"No outputs in response")
    
    structures = outputs[0].get('structures_with_scores', [])
    if not structures:
        raise Exception(f"No structures in response")
    
    # Get the first (best) structure
    pdb_content = structures[0].get('structure', '')
    if not pdb_content:
        raise Exception(f"No structure content in response")
    
    # Save predicted structure
    pdb_id = input_data.get('name', 'prediction')
    pred_path = output_dir / f"{pdb_id}_pred.pdb"
    
    with open(pred_path, 'w') as f:
        f.write(pdb_content)
    
    return pred_path, inference_time

def compare_structures(pred_pdb, reference_cif, output_dir):
    """
    Compare predicted structure against reference using OpenStructure.
    
    Args:
        pred_pdb: Path to predicted structure (PDB format)
        reference_cif: Path to reference structure (CIF format)
        output_dir: Directory to save comparison results
    
    Returns:
        dict: Comparison metrics including lDDT score
    """
    output_dir = Path(output_dir)
    output_dir.mkdir(parents=True, exist_ok=True)
    
    out_path = output_dir / "comparison_results.json"
    
    # Build OpenStructure comparison command
    cmd = [
        "compare-structures",
        "-m", str(pred_pdb),           # Model (predicted structure)
        "-r", str(reference_cif),      # Reference (ground truth)
        "--fault-tolerant",            # Handle minor structural differences
        "--min-pep-length", "4",       # Minimum peptide chain length
        "--min-nuc-length", "4",       # Minimum nucleotide chain length
        "-o", str(out_path),           # Output JSON file
        "--lddt",                      # Calculate lDDT metric
    ]
    
    # Run comparison using OpenStructure Docker container
    docker_cmd = [
        "docker", "run", "--rm",
        "-v", f"{Path.cwd()}:/home",
        "registry.scicore.unibas.ch/schwede/openstructure:latest"
    ] + cmd
    
    result = subprocess.run(
        docker_cmd,
        capture_output=True,
        text=True
    )
    
    if result.returncode != 0:
        raise Exception(f"Structure comparison failed: {result.stderr}")
    
    # Read comparison results
    with open(out_path, 'r') as f:
        metrics = json.load(f)
    
    return metrics

def extract_lddt(comparison_results):
    """
    Extract lDDT score from comparison results.
    
    Args:
        comparison_results: Dictionary containing comparison metrics
    
    Returns:
        float: lDDT score (0.0 to 1.0, higher is better)
    """
    # lDDT is directly a float value in the comparison results
    lddt_score = comparison_results.get('lddt', 0.0)
    return lddt_score

def benchmark_structure(nim_url, input_json, reference_cif, output_dir):
    """
    Complete benchmark pipeline for a single structure.
    
    Args:
        nim_url: URL of OpenFold3 NIM service
        input_json: Input configuration for prediction
        reference_cif: Reference structure for validation
        output_dir: Output directory for results
    
    Returns:
        dict: Benchmark results including timing and accuracy
    """
    output_dir = Path(output_dir)
    
    print(f"Running inference...")
    pred_path, inference_time = run_inference(nim_url, input_json, output_dir)
    print(f"Inference completed in {inference_time:.2f} seconds")
    
    print(f"Comparing structures...")
    metrics = compare_structures(pred_path, reference_cif, output_dir)
    lddt_score = extract_lddt(metrics)
    print(f"lDDT score: {lddt_score:.4f}")
    
    return {
        'inference_time': inference_time,
        'lddt_score': lddt_score,
        'predicted_structure': str(pred_path),
        'full_metrics': metrics
    }

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='Benchmark OpenFold3 NIM predictions')
    parser.add_argument('--nim-url', default='http://localhost:8000', 
                        help='URL of the OpenFold3 NIM service')
    parser.add_argument('--input', required=True, 
                        help='Path to input JSON file')
    parser.add_argument('--reference', required=True, 
                        help='Path to reference CIF file')
    parser.add_argument('--output', required=True, 
                        help='Output directory for results')
    
    args = parser.parse_args()
    
    results = benchmark_structure(args.nim_url, args.input, args.reference, args.output)
    
    print("\nBenchmark Results:")
    print(f"  Inference Time: {results['inference_time']:.2f}s")
    print(f"  lDDT Score: {results['lddt_score']:.4f}")
    print(f"  Predicted Structure: {results['predicted_structure']}")

Running a Single Benchmark

To run a benchmark for a single structure, follow these steps in order:

# Step 1: Install dependencies
pip3 install requests

# Step 2: Create directories
mkdir -p inputs results

# Step 3: Generate the input JSON file
python3 generate_inputs.py --pdb 8eil

# Step 4: Verify files exist before running benchmark
ls -l inputs/8eil_input.json references/8eil.cif

# Step 5: Run the benchmark (requires OpenFold3 NIM to be running)
python3 benchmark_openfold3.py \
    --nim-url http://localhost:8000 \
    --input inputs/8eil_input.json \
    --reference references/8eil.cif \
    --output results/8eil

Note

Each step must complete successfully before proceeding to the next. Verify that input and reference files exist (Step 4) before running the benchmark (Step 5).

Understanding lDDT Metric

lDDT (local Distance Difference Test) is a robust metric for assessing the quality of predicted protein structures:

• Score Range: 0.0 to 1.0 (or 0 to 100 when expressed as percentage)

• Interpretation:

  • lDDT > 0.90: Excellent accuracy, very high confidence

  • lDDT 0.70-0.90: Good accuracy, reliable predictions

  • lDDT 0.50-0.70: Moderate accuracy, some structural features correct

  • lDDT < 0.50: Low accuracy, limited reliability

How lDDT Works:

• Measures local geometric agreement between predicted and reference structures

• Evaluates distances between atoms within local neighborhoods (typically 15Å radius)

• More robust to domain movements and flexible regions compared to global metrics like RMSD

• Focuses on local structural correctness rather than global superposition

Why lDDT for OpenFold3:

• AlphaFold3 and OpenFold3 models are trained to optimize lDDT during the training process

• Well-suited for evaluating multi-chain complexes and structures with flexible regions

• Provides per-residue metrics in addition to global metrics

• Standard metric used in CASP (Critical Assessment of protein Structure Prediction) competitions

Running Benchmarks for Performance Table

To reproduce the performance metrics from the table above, follow these steps in order:

Important

Complete Steps 1 and 2 before running Step 3. The benchmark workflow requires:

1. Both Python scripts (generate_inputs.py and benchmark_openfold3.py) saved as files

2. The bash script (run_benchmarks.sh) saved as a file

3. All scripts must be in the same directory before execution

Running steps out of order results in “File Not Found” errors.

Step 1: Prepare the Python scripts

Save the two Python scripts provided earlier in this document:

• generate_inputs.py - Creates input JSON files from PDB (refer to Preparing Input JSON Files)

• benchmark_openfold3.py - Runs inference and compares structures (refer to Benchmarking Script)

Step 2: Create the benchmark workflow script

Save the following as run_benchmarks.sh:

#!/bin/bash
# Complete benchmark workflow for all test cases

NIM_URL="http://localhost:8000"
PDB_IDS=("8eil" "7r6r" "1a3n" "8c4d" "7qsj" "8cpk" "8are" "8owf" "8aw3" "7tpu" "7ylz" "8gpp" "8clz" "8k7x" "8ibx" "8gi1" "8sm6" "8pso" "8jue" "8bsh" "5xgo")

# Create directories
mkdir -p references inputs results

# Step 1: Generate input JSON files from PDB
echo "Step 1: Generating input JSON files..."
python3 generate_inputs.py
echo "✓ Input files created in inputs/"

# Step 2: Download reference structures
echo "Step 2: Downloading reference structures..."
for pdb_id in "${PDB_IDS[@]}"; do
    if [ ! -f "references/${pdb_id}.cif" ]; then
        echo "  Downloading ${pdb_id}.cif..."
        wget -q "https://files.rcsb.org/download/${pdb_id}.cif" -O "references/${pdb_id}.cif"
    fi
done
echo "✓ Reference structures downloaded"

# Step 3: Verify NIM is running
echo "Step 3: Checking NIM availability..."
curl -sf "${NIM_URL}/v1/health/ready" > /dev/null || {
    echo "Error: NIM not accessible at ${NIM_URL}"
    exit 1
}
echo "✓ NIM is ready"

# Step 4: Run benchmarks
echo "Step 4: Running benchmarks..."
for pdb_id in "${PDB_IDS[@]}"; do
    echo "  Benchmarking ${pdb_id}..."
    python3 benchmark_openfold3.py \
        --nim-url "$NIM_URL" \
        --input "inputs/${pdb_id}_input.json" \
        --reference "references/${pdb_id}.cif" \
        --output "results/${pdb_id}"
done

echo ""
echo "Benchmarking complete! Results saved in results/"

Step 3: Run the complete workflow

# 1. Install dependencies
pip3 install requests

# 2. Make the benchmark script executable
chmod +x run_benchmarks.sh

# 3. Run the complete workflow (generates inputs, downloads references, and runs benchmarks)
./run_benchmarks.sh

Note

The run_benchmarks.sh script handles all steps automatically: generating input files, downloading reference structures, and running benchmarks for all test cases. Make sure the OpenFold3 NIM is running before executing the script.

Expected Output Format

The comparison results JSON file contains detailed metrics:

{
  "lddt": 0.9234,
  "chain_mapping": {
    "A": "A",
    "B": "B"
  },
  "aln": [
    ">reference:A\nMKQLYGHSTI...",
    ">model:A\nMKQLYGHSTI..."
  ],
  "model_clashes": [],
  "model_bad_bonds": [],
  "model_bad_angles": [],
  "reference_clashes": [],
  "reference_bad_bonds": [],
  "reference_bad_angles": [],
  "status": "SUCCESS",
  "ost_version": "2.11.1"
}

The primary metric of interest is lddt, which is a float value (0.0 to 1.0) representing the overall structural quality of the prediction. Higher values indicate better agreement with the reference structure.

Troubleshooting

General Performance Issues

• Out of memory errors: Reduce MSA depth, decrease sequence length, or upgrade to GPUs with more memory

• Slow performance: Ensure fast storage (NVMe SSD), sufficient CPU cores, and adequate system RAM

• Poor quality predictions: Check input sequence quality, increase MSA depth if available, or adjust diffusion parameters

Benchmarking Issues

The following are “File not found” errors.

• “Input file not found”: Generate input JSON files first using generate_inputs.py

  python3 generate_inputs.py --all  # Creates all input files
  python3 generate_inputs.py --pdb 8eil  # Or for specific PDB
  

• “Reference file not found”: Download CIF files from RCSB PDB

  wget https://files.rcsb.org/download/8eil.cif -O references/8eil.cif
  

The following are structure comparison issues.

• OpenStructure comparison failures: Ensure both predicted and reference structures have compatible chain IDs and residue numbering

• Missing atoms in predictions: Use --fault-tolerant flag to handle incomplete structures (already included in script)

• lDDT score of 0.0: Check that sequences match between prediction and reference; may indicate alignment failure

The following are NIM connection issues.

• “Connection refused”: Ensure OpenFold3 NIM is running

  curl http://localhost:8000/v1/health/ready  # Should return {"object":"health.response","message":"ready","status":"ready"}
  

Note

For detailed performance tuning guidance specific to your deployment, refer to the documentation on configuration and optimization.