Merge branch '306-removeNeoAg' into 'develop'

Resolve "Remove NeoAg from NeoFox"

See merge request tron/addannot!258

MANIFEST.in

@@ -1,4 +1,3 @@
-include neofox/published_features/neoag/neoag-master/*
 include neofox/published_features/self_similarity/BLOSUM62-2.matrix.txt
 include neofox/published_features/Tcell_predictor/amino-acids-features.pickle
 include neofox/published_features/Tcell_predictor/genes-expression.pickle

README.md

@@ -42,7 +42,6 @@ NeoFox covers the following neoantigen features and prediction algorithms:
 | Vaxrank                                                 | Rubinsteyn, 2017, Front Immunol                                          | https://doi.org/10.3389/fimmu.2017.01807                                                  |
 | Priority score                                          | Bjerregaard et al, 2017, Cancer Immunol Immunother.                      | https://doi.org/10.1007/s00262-017-2001-3                                                 |
 | Tcell predictor                                         | Besser et al, 2019, Journal for ImmunoTherapy of Cancer                  | https://doi.org/10.1186/s40425-019-0595-z                                                 |
-| neoag                                                   | Smith et al, 2019, Cancer Immunology Research                            | https://doi.org/10.1158/2326-6066.CIR-19-0155                                             |
 | PRIME  §                                                 | Schmidt et al., 2021, Cell Reports Medicine                            | https://doi.org/10.1016/j.xcrm.2021.100194                                             |
 | HEX §                                                   | Chiaro et al., 2021, Cancer Immunology Research                            | https://doi.org/10.1158/2326-6066.CIR-20-0814                                             |
 

docs/source/01_overview.md

@@ -43,7 +43,6 @@ A list of implemented features and their references are given in Table 1. Please
 | Vaxrank                                                 | Rubinsteyn, 2017, Front Immunol                                          | https://doi.org/10.3389/fimmu.2017.01807                                                  |
 | Priority score                                          | Bjerregaard et al., 2017, Cancer Immunol Immunother.                      | https://doi.org/10.1007/s00262-017-2001-3                                                 |
 | Tcell predictor                                         | Besser et al., 2019, Journal for ImmunoTherapy of Cancer                  | https://doi.org/10.1186/s40425-019-0595-z                                                 |
-| neoag                                                   | Smith et al., 2019, Cancer Immunology Research                            | https://doi.org/10.1158/2326-6066.CIR-19-0155                                             |
 | PRIME v2.0 §                                                  | Schmidt et al., 2021, Cell Reports Medicine                            | https://doi.org/10.1016/j.xcrm.2021.100194                                             |
 | HEX §                                                  | Chiaro et al., 2021, Cancer Immunology Research                            | https://doi.org/10.1158/2326-6066.CIR-20-0814                                             | 
 

docs/source/03_02_output_data.md

@@ -104,7 +104,6 @@ The following table describes each of the annotations in the output:
 | Priority_score_fromRNA                      | combinatorial score of several features such as MHC binding, transcription expression and VAF in RNA                                                                                                                                          | Priority score                    |
 | Priority_score_imputed_fromDNA              | combinatorial score of several features such as MHC binding, imputed gene expression and VAF  in DNA                                                                                                                                          | Priority score                    |
 | Priority_score_imputed_fromRNA              | combinatorial score of several features such as MHC binding, imputed gene expression and VAF  in RNA                                                                                                                                         | Priority score                    |
-| Neoag_immunogenicity                        | output score of neoag model                                                                                                                                                                                                             | neoag                             |
 | IEDB_Immunogenicity_MHCI                    | IEDB Immunogenicity score for `NetMHCpan_bestAffinity_peptide`                                                                                                                                                                          | IEDB Immunogenicity               |
 | IEDB_Immunogenicity_MHCII                   | IEDB Immunogenicity score for `NetMHCIIpan_bestAffinity_peptide`                                                                                                                                                                        | IEDB Immunogenicity               |
 | MixMHCpred_bestScore_peptide                | MHC class I neoepitope candidate sequence with maximum MixMHCpred score over all neoepitope canidates (8-11mers) and MHC I alleles                                                                                                      | MixMHCpred                        |

neofox/MHC_predictors/MixMHCpred/mixmhc2pred.py

@@ -23,7 +23,7 @@
 from neofox.helpers.epitope_helper import EpitopeHelper
 from neofox.model.mhc_parser import MhcParser, get_alleles_by_gene
 
-from neofox.references.references import DependenciesConfiguration
+from neofox.references.references import DependenciesConfiguration, MhcDatabase
 
 from neofox.helpers.runner import Runner
 
@@ -46,23 +46,34 @@ class MixMHC2pred:
     ANNOTATION_PREFIX = 'MixMHC2pred'
     ANNOTATION_PREFIX_WT = 'MixMHC2pred_WT'
 
-    def __init__(self, runner: Runner, configuration: DependenciesConfiguration, mhc_parser: MhcParser):
+    def __init__(self, runner: Runner, configuration: DependenciesConfiguration, mhc_parser: MhcParser,
+                 mhc_database: MhcDatabase):
         self.runner = runner
         self.configuration = configuration
-        self.available_alleles = self._load_available_alleles()
+        self.mhc_database = mhc_database
         self.mhc_parser = mhc_parser
+        self.available_alleles = self._load_available_alleles(mhc_database)
 
         self.results = None
 
-    def _load_available_alleles(self):
+    def _load_available_alleles(self, mhc_database):
         """
         loads file with available HLA II alllels for MixMHC2pred prediction, returns set
         :return:
         """
-
-        alleles = pd.read_csv(
-            self.configuration.mix_mhc2_pred_alleles_list, skiprows=2, sep="\t"
-        )
+        if mhc_database.is_homo_sapiens():
+            alleles = pd.read_csv(
+                self.configuration.mix_mhc2_pred_human_alleles_list, skiprows=2, sep="\t"
+            )
+        # run only
+        else:
+            # to test if the required PWMdef folder for mouse is downloaded
+            if self.configuration.mix_mhc2_pred_mouse_alleles_list is not None:
+                alleles = pd.read_csv(
+                    self.configuration.mix_mhc2_pred_mouse_alleles_list, skiprows=2, sep="\t"
+                )
+            else:
+                logger.warning("The PWMdef folder of mouse has not been downloaded.")
         return list(alleles["AlleleName"])
 
 
@@ -87,7 +98,8 @@ def _combine_dq_dp_alleles(alpha_alleles: List[str], beta_alleles: List[str]):
         return alleles_pairs + alleles_triplets
 
     @staticmethod
-    def _get_mixmhc2_allele_representation(hla_alleles: List[MhcAllele]):
+    def _get_mixmhc2_allele_human_representation(hla_alleles: List[MhcAllele]):
+        # alleles: hla_alleles
         return list(
             map(
                 lambda x: "{gene}_{group}_{protein}".format(
@@ -98,12 +110,12 @@ def _get_mixmhc2_allele_representation(hla_alleles: List[MhcAllele]):
         )
 
     @staticmethod
-    def _get_mixmhc2_isoform_representation(isoform: Mhc2Isoform):
+    def _get_mixmhc2_isoform_human_representation(isoform: Mhc2Isoform):
 
-        beta_chain = MixMHC2pred._get_mixmhc2_allele_representation([isoform.beta_chain])[0]
+        beta_chain = MixMHC2pred._get_mixmhc2_allele_human_representation([isoform.beta_chain])[0]
         if isoform.alpha_chain is not None and isoform.alpha_chain.name:
             # for DR only beta chain is provided
-            alpha_chain = MixMHC2pred._get_mixmhc2_allele_representation([isoform.alpha_chain])[0]
+            alpha_chain = MixMHC2pred._get_mixmhc2_allele_human_representation([isoform.alpha_chain])[0]
             return "{alpha}__{beta}".format(alpha=alpha_chain, beta=beta_chain)
         return beta_chain
 
@@ -118,22 +130,49 @@ def transform_hla_ii_alleles_for_prediction(self, mhc: List[Mhc2]) -> List[str]:
         dqb1_alleles = get_alleles_by_gene(mhc, Mhc2GeneName.DQB1)
 
         dp_allele_combinations = self._combine_dq_dp_alleles(
-            alpha_alleles=self._get_mixmhc2_allele_representation(dpa1_alleles),
-            beta_alleles=self._get_mixmhc2_allele_representation(dpb1_alleles)
+            alpha_alleles=self._get_mixmhc2_allele_human_representation(dpa1_alleles),
+            beta_alleles=self._get_mixmhc2_allele_human_representation(dpb1_alleles)
         )
         dq_allele_combinations = self._combine_dq_dp_alleles(
-            alpha_alleles=self._get_mixmhc2_allele_representation(dqa1_alleles),
-            beta_alleles=self._get_mixmhc2_allele_representation(dqb1_alleles)
+            alpha_alleles=self._get_mixmhc2_allele_human_representation(dqa1_alleles),
+            beta_alleles=self._get_mixmhc2_allele_human_representation(dqb1_alleles)
         )
 
         return [
             a
-            for a in self._get_mixmhc2_allele_representation(drb1_alleles)
+            for a in self._get_mixmhc2_allele_human_representation(drb1_alleles)
             + dq_allele_combinations
             + dp_allele_combinations
             if a in self.available_alleles
         ]
 
+    @staticmethod
+    def _get_mixmhc2_allele_mouse_representation(h2_alleles: List[MhcAllele]):
+        return list(
+            map(
+                lambda x: "H2_{gene}a_{protein}__H2_{gene}b_{protein}".format(
+                    gene=x.gene[-1], protein=x.protein
+                ),
+                h2_alleles,
+            )
+        )
+
+    def _get_mixmhc2_isoform_mouse_representation(isoform: Mhc2Isoform):
+        if isoform is not None:
+            return "H2_{gene}a_{protein}__H2_{gene}b_{protein}".format(gene=isoform[-3], protein=isoform[-1])
+
+    def transform_h2_alleles_for_prediction(self, mhc:List[Mhc2]) -> List[str]:
+        """
+        prepares list of H2 alleles for prediction in required format
+        """
+
+        h2a_alleles = get_alleles_by_gene(mhc, Mhc2GeneName.H2A)
+        h2e_alleles = get_alleles_by_gene(mhc, Mhc2GeneName.H2E)
+
+        return [
+            a for i in (h2a_alleles, h2e_alleles) for a in self._get_mixmhc2_allele_mouse_representation(i)  if a in self.available_alleles
+        ]
+
     def _parse_mixmhc2pred_output(self, filename: str) -> List[PredictedEpitope]:
 
         parsed_results = []
@@ -157,10 +196,15 @@ def _parse_mixmhc2pred_output(self, filename: str) -> List[PredictedEpitope]:
         return parsed_results
 
     def _mixmhc2prediction(self, isoforms: List[str], potential_ligand_sequences: List[str]) -> List[PredictedEpitope]:
-
-
+        # TODO: define the pwm_path again because the mouse path is only defined by the config
         tmptxt = intermediate_files.create_temp_mixmhc2pred(potential_ligand_sequences, prefix="tmp_sequence_")
         outtmp = intermediate_files.create_temp_file(prefix="mixmhc2pred", suffix=".txt")
+
+        if self.mhc_database.is_homo_sapiens():
+            pwm_path = os.path.dirname(self.configuration.mix_mhc2_pred_human_alleles_list)
+        else:
+            #pwm_path = '/home/nguyenhv/code/MixMHC2pred/2.0/PWMdef/PWMdef_Mouse/' # reference folder
+            pwm_path = os.path.dirname(self.configuration.mix_mhc2_pred_mouse_alleles_list)
         cmd = [
             self.configuration.mix_mhc2_pred,
             "-a",
@@ -169,6 +213,8 @@ def _mixmhc2prediction(self, isoforms: List[str], potential_ligand_sequences: Li
             tmptxt,
             "-o",
             outtmp,
+            "-f",
+            pwm_path,
             "--no_context"
         ]
         self.runner.run_command(cmd)
@@ -187,10 +233,14 @@ def run(self, mhc: List[Mhc2], neoantigen: Neoantigen, uniprot):
         self.results = None
 
         potential_ligand_sequences = EpitopeHelper.generate_nmers(
-            neoantigen=neoantigen, lengths=[12,13, 14, 15, 16, 17, 18, 19, 20, 21], uniprot=uniprot)
+            neoantigen=neoantigen, lengths=[12, 13, 14, 15, 16, 17, 18, 19, 20, 21], uniprot=uniprot)
 
         if len(potential_ligand_sequences) > 0:
-            mhc2_alleles = self.transform_hla_ii_alleles_for_prediction(mhc)
+            if self.mhc_database.is_homo_sapiens():
+                mhc2_alleles = self.transform_hla_ii_alleles_for_prediction(mhc)
+            else:
+                mhc2_alleles = self.transform_h2_alleles_for_prediction(mhc)
+
             if len(mhc2_alleles) > 0:
                 self.results = self._mixmhc2prediction(
                     isoforms=mhc2_alleles, potential_ligand_sequences=potential_ligand_sequences)
@@ -203,7 +253,10 @@ def run_peptide(self, peptide: str, isoform: Mhc2Isoform) -> PredictedEpitope:
         Performs MixMHC2pred prediction for desired hla allele and writes result to temporary file.
         """
         result = None
-        isoform_representation = self._get_mixmhc2_isoform_representation(isoform)
+        if self.mhc_database.is_homo_sapiens():
+            isoform_representation = self._get_mixmhc2_isoform_human_representation(isoform)
+        else:
+            isoform_representation = self._get_mixmhc2_isoform_mouse_representation(isoform)
         if isoform_representation in self.available_alleles:
             results = self._mixmhc2prediction(
                 isoforms=[isoform_representation],

neofox/__init__.py

@@ -18,7 +18,7 @@
 # along with this program. If not, see <http://www.gnu.org/licenses/>.#
 
 
-VERSION = "1.1.0b28"
+VERSION = "1.1.0b29"
 
 REFERENCE_FOLDER_ENV = "NEOFOX_REFERENCE_FOLDER"
 NEOFOX_BLASTP_ENV = "NEOFOX_BLASTP"

neofox/annotator/neoantigen_annotator.py

@@ -31,7 +31,6 @@
 from neofox.model.factories import AnnotationFactory
 from neofox.model.mhc_parser import MhcParser
 from neofox.published_features.Tcell_predictor.tcellpredictor_wrapper import TcellPrediction
-from neofox.published_features.neoag.neoag_gbm_model import NeoagCalculator
 from neofox.published_features.self_similarity.self_similarity import SelfSimilarityCalculator
 from neofox.published_features.expression import Expression
 from neofox.model.neoantigen import Patient, Neoantigen, Annotations, PredictedEpitope
@@ -57,7 +56,6 @@ def __init__(self, references: ReferenceFolder, configuration: DependenciesConfi
         self.rank_mhcii_threshold = rank_mhcii_threshold
 
         # NOTE: these resources do not read any file thus can be initialised fast
-        self.neoag_calculator = NeoagCalculator(runner=self.runner, configuration=configuration)
         self.expression_calculator = Expression()
         self.mhc_database = references.get_mhc_database()
         self.mhc_parser = MhcParser.get_mhc_parser(self.mhc_database)
@@ -193,14 +191,6 @@ def get_annotated_neoantigen(self, neoantigen: Neoantigen, patient: Patient, wit
                 )
             )
 
-        # neoag immunogenicity model
-        if netmhcpan and netmhcpan.best_epitope_by_affinity:
-            neoantigen.neofox_annotations.annotations.append(
-                self.neoag_calculator.get_annotation(
-                    epitope_mhci=netmhcpan.best_epitope_by_affinity,
-                    neoantigen=neoantigen)
-            )
-
         # IEDB immunogenicity
         if self.organism == ORGANISM_HOMO_SAPIENS:
             neoantigen.neofox_annotations.annotations.extend(

neofox/annotator/neoantigen_mhc_binding_annotator.py

@@ -8,7 +8,7 @@
 from neofox.helpers.runner import Runner
 from neofox.model.mhc_parser import MhcParser
 from neofox.model.neoantigen import Neoantigen, Patient
-from neofox.references.references import DependenciesConfiguration, AvailableAlleles, ReferenceFolder, \
+from neofox.references.references import DependenciesConfiguration, AvailableAlleles, ReferenceFolder, MhcDatabase, \
     ORGANISM_HOMO_SAPIENS
 
 
@@ -56,16 +56,20 @@ def get_mhc_binding_annotations(self, neoantigen: Neoantigen, patient: Patient):
                 neoantigen,
                 patient
             )
+
+        if self.configuration.mix_mhc2_pred is not None and has_mhc2:
+            mixmhc2pred = self._run_mixmhc2pred(
+                self.runner,
+                self.configuration,
+                self.mhc_parser,
+                neoantigen,
+                patient,
+                self.mhc_database
+            )
+
         # avoids running MixMHCpred and PRIME for non human organisms
         if self.organism == ORGANISM_HOMO_SAPIENS:
-            if self.configuration.mix_mhc2_pred is not None and has_mhc2:
-                mixmhc2pred = self._run_mixmhc2pred(
-                    self.runner,
-                    self.configuration,
-                    self.mhc_parser,
-                    neoantigen,
-                    patient,
-                )
+
             if self.configuration.mix_mhc_pred is not None and has_mhc1:
                 mixmhcpred = self._run_mixmhcpred(
                     self.runner,
@@ -155,7 +159,8 @@ def _run_mixmhc2pred(
             mhc_parser: MhcParser,
             neoantigen: Neoantigen,
             patient: Patient,
+            mhc_database: MhcDatabase
     ):
-        mixmhc2 = MixMHC2pred(runner, configuration, mhc_parser)
+        mixmhc2 = MixMHC2pred(runner, configuration, mhc_parser, mhc_database)
         mixmhc2.run(mhc=patient.mhc2, neoantigen=neoantigen, uniprot=self.uniprot)
         return mixmhc2

neofox/annotator/neoepitope_annotator.py

@@ -32,7 +32,6 @@
 from neofox.model.factories import AnnotationFactory
 from neofox.model.mhc_parser import MhcParser
 from neofox.published_features.Tcell_predictor.tcellpredictor_wrapper import TcellPrediction
-from neofox.published_features.neoag.neoag_gbm_model import NeoagCalculator
 from neofox.published_features.self_similarity.self_similarity import SelfSimilarityCalculator
 from neofox.published_features.expression import Expression
 from neofox.model.neoantigen import Patient, Neoantigen, Annotations, PredictedEpitope
@@ -54,7 +53,6 @@ def __init__(self, references: ReferenceFolder, configuration: DependenciesConfi
         self.available_alleles = references.get_available_alleles()
 
         # NOTE: these resources do not read any file thus can be initialised fast
-        self.neoag_calculator = NeoagCalculator(runner=self.runner, configuration=configuration)
         self.mhc_database = references.get_mhc_database()
         self.mhc_parser = MhcParser.get_mhc_parser(self.mhc_database)