0. 说明：

Foldseek 是由韩国国立首尔大学 (Seoul National University) 的 Martin Steinegger (MMseqs2 和 Linclust 的作者) 开发的一款用于快速地从大型蛋白质结构数据库中检索相似结构蛋白质的工具，可以用于计算两个蛋白之间的结构相似性，可以用于蛋白质结构比对，也可以与 MMseqs2 和 Linclust 结合实现基于结构对蛋白质聚类。

本文的目的在于利用 Foldseek 将蛋白质的PDB结构转化为 3Di alphabet 表示的 3Di 序列，同时获取蛋白质蓄力的 3Di Embedding Matrix。

1. 下载和安装 Foldseek:

根据 github 上提供的安装教程（https://github.com/steineggerlab/foldseek），首先确定 Linux 的架构，然后根据相应的下载和安装命令进行 Foldseek 的下载和安装。

# Linux AVX2 build (check using: cat /proc/cpuinfo | grep avx2)
wget https://mmseqs.com/foldseek/foldseek-linux-avx2.tar.gz; tar xvzf foldseek-linux-avx2.tar.gz; export PATH=$(pwd)/foldseek/bin/:$PATH# Linux SSE2 build (check using: cat /proc/cpuinfo | grep sse2)
wget https://mmseqs.com/foldseek/foldseek-linux-sse2.tar.gz; tar xvzf foldseek-linux-sse2.tar.gz; export PATH=$(pwd)/foldseek/bin/:$PATH# Linux ARM64 build
wget https://mmseqs.com/foldseek/foldseek-linux-arm64.tar.gz; tar xvzf foldseek-linux-arm64.tar.gz; export PATH=$(pwd)/foldseek/bin/:$PATH# MacOS
wget https://mmseqs.com/foldseek/foldseek-osx-universal.tar.gz; tar xvzf foldseek-osx-universal.tar.gz; export PATH=$(pwd)/foldseek/bin/:$PATH# Conda installer (Linux and macOS)
conda install -c conda-forge -c bioconda foldseek

2. 利用 Foldseek 将PDB转化为3Di

在 foldseek 程序所在的目录下，运行：
命令：./foldseek structureto3didescriptor --help

usage: foldseek structureto3didescriptor <i:PDB|mmCIF[.gz]> ... <i:PDB|mmCIF[.gz]> <o:3didescriptor> [options]By Martin Steinegger <martin.steinegger@snu.ac.kr>
options: misc:                         --mask-bfactor-threshold FLOAT mask residues for seeding if b-factor < thr [0,100] [0.000]--file-include STR             Include file names based on this regex [.*]--file-exclude STR             Exclude file names based on this regex [^$]
common:                       --threads INT                  Number of CPU-cores used (all by default) [40]-v INT                         Verbosity level: 0: quiet, 1: +errors, 2: +warnings, 3: +info [3]
expert:                       --chain-name-mode INT          Add chain to name:0: auto1: always add[0]--write-mapping INT            write _mapping file containing mapping from internal id to taxonomic identifier [0]--coord-store-mode INT         Coordinate storage mode: 1: C-alpha as float2: C-alpha as difference (uint16_t) [2]--write-lookup INT             write .lookup file containing mapping from internal id, fasta id and file number [1]--tar-include STR              Include file names based on this regex [.*]--tar-exclude STR              Exclude file names based on this regex [^$]examples:Convert PDB/mmCIF/tar[.gz] files to a dbreferences:- van Kempen, M., Kim, S.S., Tumescheit, C., Mirdita, M., Lee, J., Gilchrist, C.L.M., Söding, J., and Steinegger, M. Fast and accurate protein structure search with Foldseek. Nature Biotechnology, doi:10.1038/s41587-023-01773-0 (2023)

从上述帮助文档的结果可以看出，将一个蛋白PDB转化为3Di，命令：
./foldseek structureto3didescriptor prot.pdb res_prot.3di --threads 1 (用一个线程将 prot.pdb 转化为 res_prot.3di)

结果示例：
将人类蛋白 A1IGU5.pdb 转化为 A1IGU5.3di，部分结果如下：

3. 从 3Di 结果中将 3Di Token 和 3Di Embedding 取出

import numpy as npdef deal3DiRes(threeDifile):with open(threeDifile) as inF:for line in inF:line = line.strip().split("\t")## 3Di Tokentoken_3di = line[-2].strip()## 3Di Embeddingmatrix_3di = np.array(line[-1].strip().split(","), dtype=float)matrix_3di_reshape = matrix_3di.reshape(-1,10) ## 因为每个氨基酸是用长度为 10 的向量来表示的，所以把最后一列分成 nx10 的矩阵即可。breakreturn token_3di, matrix_3di_reshapeif __name__ == "__main__":res = deal3DiRes(threeDifile="../VirusHumanProt3DiFiles/Human3Di/A1IGU5.3di")print(res[0]) ## 3Di 序列print(res[1]) ## 3Di matrix

A1IGU5.3di 的处理结果如下

DDDDDDDDDPDPPPPVVVVVVVLLVVLLVQLVVLLVVVLVVLVVLLVLLCCVVPQLLVLVVVDDPVLSCLLCPVSVLVSVLSVVLSVQLVVLVVPSVCNLLSNLVSLVVCLVSLLVRLLRLLLSLVVNVVSLVVQVVVVVSVVSQQVSQCVSPVVCPPVRSVSSSCSSVVVLVCPLVSLVSSLVSDDCPDPSNVSSVVSSVSSVVSNVSSVLSSLLSVLLVVFLPPDPDDPVVVVVPDDPVVVVLVVQLVVLVVCCVVVVDPADDDPLVVVLVVLLVVLLVVLVVQLVVLVVVLVVLVVVLVDQPLPDDPVPPDAPVPLVSVLSVCCSVPLSVVLSVLCCVQQNVLSVVLNSVSSSVVRLVVVLSSLSSLQVVQVVCCVVPVDDDPVSVVSNVSNVSSVVSSSVVSVVSSVVSVVSVVVSVVSVVVSVVVSVVVSVVSVVVSLVPHPCSPPDPVRVVVVVVVVVVVVVVVVVVVVVVVLVPDDFPDLDDDDVPCPVQVVVVCVVANLQFKKFQRAFDDDDDDQDDGDHGGQIWGFPACADPVRHNQWTWIDSSPDIGIDGPVRIDRRDDDPPVVNVCVVVPDDDDDDDDDDDDDDDDDDDDDPPFKKFFCDWDDDDDPQADTHHHRAIKDFPACAPPVRHNQWTFIDGPNDTHIDGSVRMDTDDDDDPDDDDDDD
[[ 2.629e-316  1.156e-316  2.629e-316 ...  1.482e-323 -1.661e+0012.872e+013][ 7.838e-001  6.043e-001  7.838e-001 ...  3.854e+000  1.000e+0006.931e-001][ 6.043e-001  1.280e-001  6.043e-001 ...  3.703e+000  1.000e+0006.931e-001]...[ 3.958e-001  3.127e-001 -1.000e+000 ...  3.815e+000 -1.000e+000-6.931e-001][ 3.945e-001  3.958e-001 -1.000e+000 ...  3.853e+000 -1.000e+000-6.931e-001][ 0.000e+000  0.000e+000  0.000e+000 ...  0.000e+000  0.000e+0000.000e+000]]

参考：

[1]. van Kempen M, Kim S, Tumescheit C, Mirdita M, Lee J, Gilchrist C, Söding J, and Steinegger M. Fast and accurate protein structure search with Foldseek. Nature Biotechnology, doi:10.1038/s41587-023-01773-0 (2023)
[2]. Barrio-Hernandez I, Yeo J, Jänes J, Mirdita M, Gilchrist LMC, Wein T, Varadi M, Velankar S, Beltrao P and Steinegger M. Clustering predicted structures at the scale of the known protein universe. Nature, doi:10.1038/s41586-023-06510-w (2023)
[3]. https://github.com/steineggerlab/foldseek