spectrseqtools 0.1.0__py3-none-any.whl

spectrseqtools/assets/element_masses.tsv

@@ -0,0 +1,6 @@
+symbol	mass
+H+	1.007276
+H	1.00783
+C	12.00000
+O	15.99493
+P	30.97376

spectrseqtools/assets/elemental_composition.tsv

@@ -0,0 +1,5 @@
+base	C	H	N	O	P	S
+adenosine	10	13	5	4	0	0
+cytidine	9	13	3	5	0	0
+guanosine	10	13	5	5	0	0
+uridine	9	12	2	6	0	0

spectrseqtools/assets/masses.tsv

@@ -0,0 +1,145 @@
+nucleoside	canonical_name	monoisotopic_mass	modification_rate
+A	A	267.09675	1.0
+C	C	243.08552	1.0
+G	G	283.09167	1.0
+U	U	244.06954	1.0
+0A	Am	281.1124	1.0
+00A	Ar(p)	479.1053	1.0
+01A	m1Am	295.1281	1.0
+019A	m1Im	296.1121	1.0
+06A	m6Am	295.1281	1.0
+066A	m6,6Am	309.1437	1.0
+09A	Im	282.0964	1.0
+1A	m1A	281.1124	1.0
+19A	m1I	282.0964	1.0
+2A	m2A	281.1124	1.0
+21161A	msms2i6A	427.1347	1.0
+2160A	ms2io6A	397.142	1.0
+2161A	ms2i6A	381.1471	1.0
+2162A	ms2t6A	458.122	1.0
+2163A	ms2hn6A	472.1376	1.0
+2164A	ms2ct6A	440.1114	1.0
+2165A	ht6A	428.1292	1.0
+28A	m2,8A	295.1281	1.0
+6A	m6A	281.1124	1.0
+60A	io6A	351.1543	1.0
+61A	i6A	335.1594	1.0
+62A	t6A	412.1343	1.0
+621A	ms2m6A	327.1001	1.0
+63A	hn6A	426.1499	1.0
+64A	ac6A	309.1073	1.0
+65A	g6A	368.108	1.0
+66A	m6,6A	295.1281	1.0
+662A	m6t6A	426.1499	1.0
+67A	f6A	295.0917	1.0
+68A	hm6A	297.1073	1.0
+69A	ct6A	394.1237	1.0
+8A	m8A	281.1124	1.0
+9A	I	268.0808	1.0
+0C	Cm	257.10117055	1.0
+04C	m4Cm	271.11682061	1.0
+042C	ac4Cm	299.11173523	1.0
+044C	m4,4Cm	285.13247067	1.0
+05C	m5Cm	271.11682061	1.0
+051C	hm5Cm	287.1117	1.0
+071C	f5Cm	285.09608517	1.0
+2C	s2C	259.06267687	1.0
+20C	C+	355.1965	1.0
+21C	k2C	371.18048347	1.0
+3C	m3C	257.10117055	1.0
+4C	m4C	257.10117055	1.0
+42C	ac4C	285.09608517	1.0
+44C	m4,4C	271.11682061	1.0
+5C	m5C	257.10117055	1.0
+50C	ho5C	259.08043511	1.0
+51C	hm5C	273.09608517	1.0
+71C	f5C	271.08043511	1.0
+0G	Gm	297.1073	1.0
+00G	Gr(p)	495.1003	1.0
+01G	m1Gm	311.123	1.0
+02G	m2Gm	311.123	1.0
+022G	m2,2Gm	325.1386	1.0
+027G	m2,7Gm	327.1543	1.0
+1G	m1G	297.1073	1.0
+10G	Q	409.1597	1.0
+100G	preQ0	307.0917	1.0
+101G	preQ1	311.123	1.0
+102G	oQ	425.1547	1.0
+103G	G+	324.1182	1.0
+104G	galQ	571.2126	1.0
+105G	gluQ	538.2023	1.0
+106G	manQ	571.2126	1.0
+2G	m2G	297.1073	1.0
+22G	m2,2G	311.123	1.0
+227G	m2,2,7G	327.1543	1.0
+27G	m2,7G	313.1386	1.0
+34G	imG	335.123	1.0
+342G	mimG	349.1386	1.0
+347G	yW-72	436.1706	1.0
+3470G	OHyWx	452.1655	1.0
+348G	yW-58	450.1863	1.0
+3480G	OHyWy	466.1812	1.0
+3483G	yW	508.1918	1.0
+34830G	OHyW	524.1867	1.0
+34832G	o2yW	540.1816	1.0
+4G	imG-14	321.1073	1.0
+42G	imG2	335.123	1.0
+47G	yW-86	422.155	1.0
+7G	m7G	299.123	1.0
+0U	Um	258.08518614	1.0
+02U	s2Um	274.06234252	1.0
+03U	m3Um	272.1008362	1.0
+05U	m5Um	272.1008362	1.0
+0503U	mcmo5Um	346.10123012	1.0
+051U	cmnm5Um	345.11721453	1.0
+0521U	mcm5Um	330.1063155	1.0
+0522U	mchm5Um	346.10123012	1.0
+053U	ncm5Um	315.10664985	1.0
+0583U	inm5Um	355.17433547	1.0
+09U	Ym	258.08518614	1.0
+1309U	m1acp3Y	359.13286459	1.0
+19U	m1Y	258.08518614	1.0
+2U	s2U	260.04669246	1.0
+20U	se2U	307.9911	1.0
+2051U	cmnm5se2U	395.0232	1.0
+20510U	nm5se2U	337.0177	1.0
+20511U	mnm5se2U	365.049	1.0
+21U	ges2U	396.17189294	1.0
+2151U	cmnm5ges2U	483.20392133	1.0
+21510U	nm5ges2U	425.19844	1.0
+21511U	mnm5ges2U	439.21409209	1.0
+25U	m5s2U	274.0623	1.0
+251U	cmnm5s2U	347.07872085	1.0
+2510U	nm5s2U	289.07324155	1.0
+2511U	mnm5s2U	303.08889161	1.0
+2521U	mcm5s2U	332.06782182	1.0
+253U	ncm5s2U	317.06815617	1.0
+254U	tm5s2U	397.06135653	1.0
+2540U	cm5s2U	318.05217176	1.0
+2583U	inm5s2U	357.13584179	1.0
+3U	m3U	258.08518614	1.0
+30U	acp3U	345.11721453	1.0
+308U	acp3D	347.13286459	1.0
+309U	acp3Y	345.11721453	1.0
+39U	m3Y	258.08518614	1.0
+5U	m5U	258.08518614	1.0
+50U	ho5U	260.0644507	1.0
+501U	mo5U	274.08010076	1.0
+502U	cmo5U	318.06993	1.0
+503U	mcmo5U	332.08558006	1.0
+51U	cmnm5U	331.10156447	1.0
+510U	nm5U	273.09608517	1.0
+511U	mnm5U	287.11173523	1.0
+52U	cm5U	302.07501538	1.0
+520U	chm5U	318.06993	1.0
+521U	mcm5U	316.09066544	1.0
+522U	mchm5U	332.08558006	1.0
+53U	ncm5U	301.09099979	1.0
+531U	nchm5U	317.08591441	1.0
+54U	tm5U	381.08420015	1.0
+55U	cnm5U	283.08043511	1.0
+58U	m5D	260.1008362	1.0
+583U	inm5U	341.15868541	1.0
+74U	s4U	260.04669246	1.0
+8U	D	246.08518614	1.0
+9U	Y	244.06953608	1.0

spectrseqtools/cli.py

@@ -0,0 +1,259 @@
+import os
+import polars as pl
+import yaml
+from pathlib import Path
+from tap import Tap
+from typing import List, Literal
+
+from spectrseqtools.fragment_classification import classify_fragments
+from spectrseqtools.mass_table import DynamicProgrammingTable, SequenceInformation
+from spectrseqtools.masses import (
+    COMPRESSION_RATE,
+    DEFAULT_INTENSITY_CUTOFF,
+    EXPLANATION_MASSES,
+    MATCHING_THRESHOLD,
+    NUC_REPS,
+    TOLERANCE,
+    UNMODIFIED_BASES,
+    build_breakage_dict,
+)
+from spectrseqtools.prediction import Predictor
+from spectrseqtools.preprocessing import preprocess
+
+
+class Settings(Tap):
+    fragments: Path  # Path to TSV table or RAW data of observed fragments to use for prediction
+    meta: Path  # Path to YAML with meta information to use for prediction
+    fragment_predictions: (
+        Path  # Path to TSV table that shall contain the per fragment predictions
+    )
+    sequence_prediction: (
+        Path  # Path to FASTA file that shall contain the predicted sequence
+    )
+    output_dir: Path = None  # Output directory (default: input directory)
+    sequence_name: str
+    modification_rate: float = 0.5  # Maximum percentage of modification in sequence
+    solver: Literal["gurobi", "cbc"] = (
+        "gurobi"  # Solver to use for the optimization problem
+    )
+    lp_timeout_short: int = 5  # Time-out for shorter solving of LP instances
+    lp_timeout_long: int = 60  # Time-out for longer solving of LP instances
+    cutoff_percentile: int = 75  # Intensity percentile used as cutoff
+    threads: int = 1  # Number of threads to use for the optimization problem
+
+
+def main():
+    settings = Settings(underscores_to_dashes=True).parse_args()
+
+    # Set parameters for LP solver
+    solver_params = {
+        "fixed": {
+            "solver": select_solver(settings.solver),
+            "threads": settings.threads,
+            "msg": False,
+        },
+        "timeLimit(short)": settings.lp_timeout_short,
+        "timeLimit(long)": settings.lp_timeout_long,
+    }
+
+    settings.fragments = settings.fragments.resolve()
+    fragment_dir = (
+        settings.fragments.parent
+        if settings.output_dir is None
+        else settings.output_dir
+    )
+    file_prefix = settings.fragments.stem
+    with open(settings.meta, "r") as f:
+        meta = yaml.safe_load(f)
+
+    # Preprocess data if necessary
+    match settings.fragments.suffix:
+        case ".raw":
+            print("RAW file found. Preprocessing raw data...")
+            # Preprocess raw data
+            fragments, singletons, meta = preprocess(
+                file_path=settings.fragments,
+                deconvolution_params={},
+                meta_params=meta,
+                cutoff_percentile=settings.cutoff_percentile,
+            )
+            # Save preprocessed fragments
+            fragments.write_csv(fragment_dir / f"{file_prefix}.tsv", separator="\t")
+
+            # Save singletons detected from raw data
+            singletons.write_csv(
+                fragment_dir / f"{file_prefix}.singletons.tsv", separator="\t"
+            )
+
+            # Save updated meta data
+            with open(fragment_dir / f"{file_prefix}.preprocessed.meta.yaml", "w") as f:
+                yaml.dump(meta, f)
+
+            print("Preprocessing completed!\n")
+        case ".tsv":
+            print("TSV file found. Proceeding without preprocessing.")
+            # Read already preprocessed fragments
+            fragments = pl.read_csv(settings.fragments, separator="\t")
+
+            # Read singletons if given
+            singletons = None
+            if os.path.isfile(fragment_dir / f"{file_prefix}.singletons.tsv"):
+                singletons = pl.read_csv(
+                    fragment_dir / f"{file_prefix}.singletons.tsv", separator="\t"
+                )
+        case _:
+            raise NotImplementedError(
+                "Support is currently only given for TSV or RAW files."
+            )
+
+    print("Singletons identified during preprocessing:", singletons)
+    print()
+
+    explanation_masses = EXPLANATION_MASSES
+
+    # Filter by singletons
+    if singletons is not None:
+        # Map singletons to their mass representative
+        singletons = singletons.with_columns(
+            pl.col("nucleoside").replace_strict(NUC_REPS).alias("nucleoside")
+        )
+
+        # Select only bases found in singletons
+        explanation_masses = explanation_masses.with_columns(
+            pl.when(
+                pl.col("nucleoside").is_in(
+                    singletons.get_column("nucleoside").to_list()
+                )
+            )
+            .then(pl.col("modification_rate"))
+            .otherwise(pl.lit(0.0))
+            .alias("modification_rate")
+        )
+
+    # Ensure modification rates of unmodified bases are set to 1
+    explanation_masses = explanation_masses.with_columns(
+        pl.when(~pl.col("nucleoside").is_in(UNMODIFIED_BASES))
+        .then(pl.col("modification_rate"))
+        .otherwise(pl.lit(1.0))
+        .alias("modification_rate")
+    )
+
+    # Read additional parameter from meta file
+    intensity_cutoff = meta.setdefault("intensity_cutoff", DEFAULT_INTENSITY_CUTOFF)
+    start_tag = meta.setdefault("label_mass_5T", 555.1294)
+    end_tag = meta.setdefault("label_mass_3T", 455.1491)
+
+    # Build breakage dict
+    breakage_dict = build_breakage_dict(mass_5_prime=start_tag, mass_3_prime=end_tag)
+
+    # Standardize sequence mass (remove START_END breakage to gain SU mass)
+    seq_mass_obs = meta["sequence_mass"]
+    seq_mass_su = (
+        seq_mass_obs
+        - [
+            mass * TOLERANCE
+            for mass in breakage_dict
+            if "START_END" in breakage_dict[mass]
+        ][0]
+    )
+
+    # Initialize SequenceInformation class
+    seq_info = SequenceInformation(
+        max_len=int(
+            seq_mass_su
+            / TOLERANCE
+            / min(
+                pl.Series(
+                    explanation_masses.filter(pl.col("modification_rate") > 0.0).select(
+                        "tolerated_integer_masses"
+                    )
+                ).to_list()
+            )
+        ),
+        su_mass=seq_mass_su,
+        obs_mass=seq_mass_obs,
+        modification_rate=settings.modification_rate,
+    )
+
+    # Initialize DynamicProgrammingTable class
+    dp_table = DynamicProgrammingTable(
+        nucleotide_df=explanation_masses,
+        compression_rate=int(COMPRESSION_RATE),
+        tolerance=MATCHING_THRESHOLD,
+        precision=TOLERANCE,
+        seq=seq_info,
+    )
+
+    print("Alphabet after singleton reduction:")
+    dp_table.print_masses()
+    print()
+
+    # Classify preprocessed fragments
+    fragments = classify_fragments(
+        fragment_masses=fragments,
+        dp_table=dp_table,
+        breakage_dict=breakage_dict,
+        output_file_path=fragment_dir / f"{file_prefix}.standard_unit_fragments.tsv",
+        intensity_cutoff=intensity_cutoff,
+    )
+
+    # Predict sequence
+    prediction = Predictor(
+        dp_table=dp_table,
+        explanation_masses=explanation_masses,
+    ).predict(
+        fragments=fragments,
+        solver_params=solver_params,
+    )
+
+    print("Predicted sequence =\t", prediction.sequence)
+
+    # Save fragment predictions
+    prediction.fragments.write_csv(settings.fragment_predictions, separator="\t")
+
+    # Save predicted sequence
+    with open(settings.sequence_prediction, "w") as f:
+        print(f">{settings.sequence_name}", file=f)
+        print("".join(prediction.sequence), file=f)
+        print(f">{settings.sequence_name}_full", file=f)
+        print(format_sequence_to_full_version(seq=prediction.sequence), file=f)
+
+
+def format_sequence_to_full_version(seq: List[str]) -> str:
+    """
+    Format a sequence to its full version (i.e. include alternate nucleotides).
+
+    Parameters
+    ----------
+    seq: List[str]
+        Given predicted sequence.
+
+    Returns
+    -------
+    str
+        Sequence with all alternate nucleotides.
+
+    """
+    output = ""
+    for nuc in seq:
+        alt_nucs = (
+            EXPLANATION_MASSES.filter(pl.col("nucleoside") == nuc)
+            .select("nucleoside_list")
+            .item()
+            .to_list()
+        )
+        if len(alt_nucs) == 1:
+            output += nuc
+        else:
+            output += "[" + "|".join(alt_nucs) + "]"
+    return output
+
+
+def select_solver(solver: str):
+    match solver:
+        case "gurobi":
+            return "GUROBI_CMD"
+        case "cbc":
+            return "PULP_CBC_CMD"
+        case _:
+            raise NotImplementedError(f"Support for '{solver}' is currently not given.")

spectrseqtools/common.py

@@ -0,0 +1,93 @@
+import ms_deisotope as ms_ditp
+import re
+
+from clr_loader import get_mono
+from typing import List
+
+from spectrseqtools.mass_explanation import explain_mass_with_table
+from spectrseqtools.mass_table import DynamicProgrammingTable
+
+rt = get_mono()
+
+ERROR_METHOD = "l1_norm"
+_NUCLEOSIDE_RE = re.compile(r"\d*[ACGU]")
+
+
+def parse_nucleosides(sequence: str):
+    return _NUCLEOSIDE_RE.findall(sequence)
+
+
+class Explanation:
+    def __init__(self, *nucleosides):
+        self.nucleosides = tuple(sorted(nucleosides))
+
+    def __iter__(self):
+        yield from self.nucleosides
+
+    def __len__(self):
+        return len(self.nucleosides)
+
+    def __repr__(self):
+        return f"{{{','.join(self.nucleosides)}}}"
+
+    def __eq__(self, other):
+        return self.nucleosides == other
+
+
+def calculate_error_threshold(mass1: float, mass2: float, threshold: float) -> float:
+    match ERROR_METHOD:
+        case "l1_norm":
+            return threshold * (mass1 + mass2)
+        case "l2_norm":
+            return threshold * ((mass1**2 + mass2**2) ** 0.5)
+        case _:
+            raise NotImplementedError("This error method is not implemented.")
+
+
+def calculate_explanations(
+    diff: float,
+    threshold: float,
+    dp_table: DynamicProgrammingTable,
+) -> List[Explanation]:
+    explanation_list = explain_mass_with_table(
+        diff,
+        dp_table=dp_table,
+        max_modifications=round(dp_table.seq.modification_rate * dp_table.seq.max_len),
+        threshold=threshold,
+    ).explanations
+
+    # Return None if no explanation was found
+    if explanation_list is None:
+        return None
+
+    # Return all found explanations
+    explanation_list = list(explanation_list)
+    return [Explanation(*explanation_list[i]) for i in range(len(explanation_list))]
+
+
+def initialize_raw_file_iterator(
+    file_path: str,
+) -> ms_ditp.data_source.thermo_raw_net.ThermoRawLoader:
+    """
+    Initialize iterator over scans in ThermoFisher RAW file format.
+
+    Parameters
+    ----------
+    file_path : str
+        Path of RAW file from ThermoFisher.
+
+    Returns
+    -------
+    raw_file : ms_deisotope.data_source.thermo_raw_net.ThermoRawLoader
+        Iterator over scans from RAW file.
+
+    """
+    # Read data from file
+    raw_file = ms_ditp.data_source.thermo_raw_net.ThermoRawLoader(
+        file_path, _load_metadata=True
+    )
+
+    # Initialize an iterator while ungrouping MS1 from MS2 scans
+    raw_file.make_iterator(grouped=False)
+
+    return raw_file