You may try this

https://www.expasy.org/proteomics/protein_structure

The best source is uniprot database

http://www.uniprot.org/

Pfam database

http://pfam.xfam.org/

and other database for special requirements such as STRING database

Agree with Artur Burzynski

Main article: Protein structure prediction

IntFOLD is an integrated server for modelling protein structures and functions from amino acid sequences.[1] Each major version of the IntFOLD server has been independently tested in the CASP9-CASP12 experiments and its component algorithms have performed well, ranking among the top servers in the tertiary structure, quality assessment, disorder and function prediction categories (http://predictioncenter.org/). Over the years, the IntFOLD server has performed particularly well at self-estimates of 3D model accuracy. In the CASP12 experiment, according to the assessors formula the server ranked 3rd/5th (average Z-score > 0.0/–2.0) on the server only targets in the template based modelling category, and the McGuffin group manual target predictions ranked 1st/2nd (average Z-score > −2.0/0.0)[2][3]. The tertiary structure predictions from the server are also continuously benchmarked on CAMEO (http://www.cameo3d.org). The method is competitive and continual performance improvements have been observed in the successive versions.

I-TASSER is the best server for protein structure prediction according to the 2006–2016 CASP experiments (CASP7, CASP8, CASP9, CASP10, CASP11, and CASP12). The standalone I-TASSER package is freely available for download.

HHpred was the leading server for template-based protein structure prediction in the 2010 CASP9 experiment. It has a median response time of a few minutes instead of days like other top-performing servers. HHpred is often used for remote homology detection and homology-based function prediction. It runs with the free, open-source software package HH-suite for fast sequence searching, protein threading and remote homology detection.

RaptorX excels at aligning hard targets according to the 2010 CASP9 experiments. RaptorX generates the significantly better alignments for the hardest 50 CASP9 template-based modeling targets than other servers including those using consensus and refinement methods. The RaptorX server is available at server

MODELLER is a popular software tool for producing homology models by satisfaction of spatial restraints using methodology derived from NMR spectroscopy data processing. The ModWeb comparative protein structure modeling web-server uses primarily MODELLER for automatic comparative modeling.

Robetta from the Baker Laboratory at the University of Washington Robetta is the top-performing Homology Modeling server at CASP 2016 (Template Based Modeling) CASP12and is the top performing server in the Continuous Automated Model EvaluatiOn testing server as of June 3, 2017, and has held that number one rank at least since 2014 (CAMEO) CAMEO. Robetta is a ten-step pipeline of multiple bioinformatics tools, notably the Rosetta software suite of algorithms for computational modeling and analysis of protein structures. Rosetta development began in the laboratory of Dr. Baker as a structure prediction tool but since then has been adapted to solve common computational macromolecular problems. [4] Extensions of Rosetta include Foldit, which seeks to investigate the pattern-recognition and puzzle-solving abilities inherent to the human mind in order to create more successful computer protein structure prediction software.

TIP is a knowledgebase of STRUCTFAST models and precomputed similarity relationships between sequences, structures, and binding sites. Several distributed computingprojects concerning protein structure prediction have also been implemented, such as the Folding@home, Rosetta@home, Human Proteome Folding Project, Predictor@home, and TANPAKU.

BBSP (Building Blocks Structure Predictor) is a program that makes use of Hybrid template-based approaches, which associate fragment conformations for the sequence and detect distant fold similarities based on the fragment similarities [5]. Also, such program is assisted by a secondary structure prediction algorithm named HCAM, which is mainly based on hydropathy profile analysis and is included in the BBSP tools suite.[6]

MOE (Molecular Operating Environment) is a platform for protein and small molecule modelling and drug discovery applications, including homology modelling, with custom, knowledge-based routines for different protein and antibody families, e.g. GPCRs, kinases, etc. There are also applications for analysing modelled or experimental structures, and making them suitable to use in drug discovery programs.

You can make use of it:

https://www.bbcgoodfood.com/howto/guide/best-sources-protein