1+ module TMscore
2+
3+ import TMscore_jll
4+
5+ export tmscore, run_tmscore, TMscoreResult
6+
7+ """
8+ TMscoreResult
9+
10+ Holds all of the key metrics returned by the TMscore executable.
11+ """
12+ struct TMscoreResult
13+ output:: String # raw output from TMscore
14+ len1:: Int # length of structure 1
15+ len2:: Int # length of structure 2
16+ common:: Int # number of residues in common
17+ rmsd:: Float64 # RMSD over the common residues
18+ tmscore:: Float64 # TM‐score
19+ d0:: Float64 # d0 distance
20+ maxsub:: Float64 # MaxSub‐score
21+ maxsub_d0:: Float64 # MaxSub d0
22+ gdt_ts:: Float64 # GDT‐TS score
23+ gdt_ts_thresholds:: Dict{Float64,Float64} # (d<1, d<2, d<4, d<8) → value
24+ gdt_ha:: Float64 # GDT‐HA score
25+ gdt_ha_thresholds:: Dict{Float64,Float64} # (d<0.5, d<1, d<2, d<4) → value
26+ rotation:: Matrix{Float64} # 3×3 rotation matrix
27+ translation:: Vector{Float64} # length‐3 translation vector
28+ end
29+
30+ Base. show (io:: IO , result:: TMscoreResult ) = print (io, TMscoreResult, " (\"\"\"\n " . output, " \"\"\" )"
31+
32+ """
33+ TMscoreResult(output::String)::TMscoreResult
34+
35+ Parse the text printed by TMscore into a `TMscoreResult`.
36+ """
37+ function TMscoreResult (output:: String ):: TMscoreResult
38+ len1 = 0 ; len2 = 0 ; common = 0 ; rmsd = 0.0
39+ tms = 0.0 ; d0 = 0.0
40+ ms = 0.0 ; ms_d0 = 0.0
41+ gdt_ts = 0.0 ; gdt_ha = 0.0
42+ gdt_ts_th = Dict {Float64,Float64} (); gdt_ha_th = Dict {Float64,Float64} ()
43+ rotation = zeros (Float64, 3 , 3 )
44+ translation = zeros (Float64, 3 )
45+
46+ for line in split (output, ' \n '
47+ s = strip (line)
48+
49+ # Structure lengths
50+ if startswith (s, " Structure1:"
51+ m = match (r" Length=\s *(\d +)"
52+ if m != = nothing
53+ len1 = parse (Int, m. captures[1 ])
54+ end
55+ elseif startswith (s, " Structure2:"
56+ m = match (r" Length=\s *(\d +)"
57+ if m != = nothing
58+ len2 = parse (Int, m. captures[1 ])
59+ end
60+
61+ # common residues & RMSD
62+ elseif startswith (s, " Number of residues in common="
63+ m = match (r" common=\s *(\d +)"
64+ if m != = nothing
65+ common = parse (Int, m. captures[1 ])
66+ end
67+ elseif startswith (s, " RMSD"
68+ m = match (r" =\s *([\d\. ]+)"
69+ if m != = nothing
70+ rmsd = parse (Float64, m. captures[1 ])
71+ end
72+
73+ # TM-score + d0
74+ elseif startswith (s, " TM-score"
75+ m = match (r" TM-score\s *=\s *([\d\. ]+).*\( d0=\s *([\d\. ]+)\) "
76+ if m != = nothing
77+ tms = parse (Float64, m. captures[1 ])
78+ d0 = parse (Float64, m. captures[2 ])
79+ end
80+
81+ # MaxSub-score + d0
82+ elseif startswith (s, " MaxSub-score"
83+ m = match (r" MaxSub-score\s *=\s *([\d\. ]+).*\( d0=\s *([\d\. ]+)\) "
84+ if m != = nothing
85+ ms = parse (Float64, m. captures[1 ])
86+ ms_d0 = parse (Float64, m. captures[2 ])
87+ end
88+
89+ # GDT-TS
90+ elseif startswith (s, " GDT-TS-score"
91+ m = match (r" =\s *([\d\. ]+)"
92+ if m != = nothing
93+ gdt_ts = parse (Float64, m. captures[1 ])
94+ end
95+ for mm in eachmatch (r" \( d<([\d\. ]+)\) =([\d\. ]+)"
96+ thr = parse (Float64, mm. captures[1 ])
97+ val = parse (Float64, mm. captures[2 ])
98+ gdt_ts_th[thr] = val
99+ end
100+
101+ # GDT-HA
102+ elseif startswith (s, " GDT-HA-score"
103+ m = match (r" =\s *([\d\. ]+)"
104+ if m != = nothing
105+ gdt_ha = parse (Float64, m. captures[1 ])
106+ end
107+ for mm in eachmatch (r" \( d<([\d\. ]+)\) =([\d\. ]+)"
108+ thr = parse (Float64, mm. captures[1 ])
109+ val = parse (Float64, mm. captures[2 ])
110+ gdt_ha_th[thr] = val
111+ end
112+
113+ end
114+
115+ # rotation/translation rows
116+ m = match (r" ^\s *(\d +)\s +([-\d\. ]+)\s +([-\d\. ]+)\s +([-\d\. ]+)\s +([-\d\. ]+)"
117+ if m != = nothing
118+ i = parse (Int, m. captures[1 ])
119+ translation[i] = parse (Float64, m. captures[2 ])
120+ rotation[i, 1 : 3 ] .= parse .(Float64, m. captures[3 : 5 ])
121+ end
122+ end
123+
124+ return TMscoreResult (
125+ output,
126+ len1, len2, common, rmsd,
127+ tms, d0, ms, ms_d0,
128+ gdt_ts, gdt_ts_th, gdt_ha, gdt_ha_th,
129+ rotation, translation
130+ )
131+ end
132+
133+ tmscore_command (options:: Vector{String} ) = Cmd ([TMscore_jll. TMscore (). exec; options])
134+
135+ """
136+ run_tmscore(file1::String, file2::String; options...) -> TMscoreResult
137+ run_tmscore(struc1, struc2; options...) -> TMscoreResult
138+
139+ Invoke the TMscore binary and parse its output into a `TMscoreResult` struct.
140+
141+ `BioStructures.StructuralElementOrList` objects can also be used as input,
142+ and will be written to a temporary file before processing.
143+
144+ ## Options
145+
146+ ### Boolean Flags
147+ - `c::Bool`: Compare two complex structures with multiple chains
148+ - `seq::Bool`: Establish residue equivalence by sequence alignment instead of residue indices
149+ - `a::Bool`: TM-score normalized by the average length of two structures
150+ - `m::Bool`: Output TM-score rotation matrix
151+ - `fast::Bool`: Fast but slightly inaccurate alignment
152+ - `mirror::Bool`: Whether to align the mirror image of input structure
153+ - `het::Bool`: Whether to align residues marked as 'HETATM' in addition to 'ATOM'
154+
155+ ### Numeric Options
156+ - `d::Number`: TM-score scaled by an assigned d0 (in Angstroms)
157+ - `l::Int`: TM-score normalized by a specific length
158+ - `ter::Int`: Strings to mark the end of a chain
159+ - 0: No TER card
160+ - 1: TER separates different chains
161+ - 2: TER marks end of each chain
162+ - 3: EXIT separates different chains
163+ - `split::Int`: Whether to split PDB file into multiple chains
164+ - 0: Don't split
165+ - 1: Split by chain ID
166+ - 2: Split by TER records
167+ - `outfmt::Int`: Output format
168+ - 0: Full format
169+ - 1: Sequence and structure in fasta format
170+ - 2: Matrix format
171+ - -1: Compact format
172+ - `infmt1::Int`: Input format for chain1
173+ - -1: Auto-detect
174+ - 0: PDB format
175+ - 1: SPICKER format
176+ - 2: xyz format
177+ - 3: FASTA format
178+ - `infmt2::Int`: Input format for chain2 (same options as infmt1)
179+
180+ ### String Options
181+ - `o::String`: Generate superposition output files with the given prefix
182+ - `dir::String`: Perform all-against-all alignment among the list of PDB chains
183+ - `dir1::String`: Use chain2 to search a list of PDB chains
184+ - `dir2::String`: Use chain1 to search a list of PDB chains
185+ - `suffix::String`: Add file name suffix to files listed by chain1_list or chain2_list
186+ - `atom::String`: 4-character atom name used to represent a residue
187+ - `mol::String`: Molecule type: RNA or protein
188+ """
189+ function run_tmscore (file1:: AbstractString , file2:: AbstractString ; options... )
190+ cmd_vec = String[file1, file2]
191+ for (key, val) in pairs (options)
192+ if val === true
193+ push! (cmd_vec, " -$key "
194+ elseif val === false
195+ nothing
196+ else
197+ push! (cmd_vec, " -$key " string (val))
198+ end
199+ end
200+
201+ cmd = tmscore_command (cmd_vec)
202+ stdout_buffer = IOBuffer ()
203+ stderr_buffer = IOBuffer ()
204+ success = try
205+ process = run (pipeline (cmd, stdout = stdout_buffer, stderr = stderr_buffer); wait= true )
206+ process. exitcode == 0
207+ catch e
208+ false
209+ end
210+ stdout_output = String (take! (stdout_buffer))
211+ stderr_output = String (take! (stderr_buffer))
212+ if success
213+ return TMscoreResult (stdout_output)
214+ else
215+ error_msg = " TMscore failed with error:\n " *
216+ (isempty (stderr_output) ? stdout_output : stderr_output)
217+ throw (ErrorException (error_msg))
218+ end
219+ end
220+
221+ write_tempfile (filename:: AbstractString , tempdir) = filename
222+
223+ function run_tmscore (arg1, arg2; options... )
224+ mktempdir () do tempdir
225+ run_tmscore (write_tempfile (arg1, tempdir), write_tempfile (arg2, tempdir); options... )
226+ end
227+ end
228+
229+ """
230+ tmscore(file1::String, file2::String; options...) -> Float64
231+ tmscore(struc1, struc2; options...) -> Float64
232+
233+ Invoke the TMscore binary and return only the TM-score value.
234+
235+ This is a convenience function that calls `run_tmscore` internally and extracts
236+ the `tmscore` field from the resulting `TMscoreResult`. See `run_tmscore` for
237+ details on options and handling different input types.
238+ """
239+ function tmscore (arg1, arg2; options... )
240+ result = run_tmscore (arg1, arg2; options... )
241+ return result. tmscore
242+ end
243+
244+
245+ # # BioStructuresExt.jl
246+
247+ using BioStructures
248+
249+ function TMscore. write_tempfile (arg:: StructuralElementOrList , tempdir)
250+ path = joinpath (tempdir, " $(time_ns ()) .cif"
251+ writemmcif (path, arg)
252+ return path
253+ end
254+
255+ end
0 commit comments