Prosit ptm

.github/workflows/mergeCi.yaml

@@ -14,8 +14,11 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v4
 
-      - name: Startup
-        run: docker compose --env-file .env.ci up -d --build serving develop --wait
+      - name: Startup develop
+        run: docker compose --env-file .env.ci up -d --build develop --wait
+
+      - name: Startup Serving
+        run: docker compose --env-file .env.ci up -d --build serving --wait
 
       - name: Styling
         run: docker compose --env-file .env.ci exec develop koina_format.sh --check

Dockerfile

@@ -1,6 +1,6 @@
 FROM nvcr.io/nvidia/tritonserver:23.05-py3 AS serving-develop
 RUN pip install requests ms2pip==3.13 psm-utils pandas pyteomics==4.6.2 rdkit==2024.3.5
-HEALTHCHECK --start-period=10m --interval=15s --retries=1 CMD curl --fail localhost:8501/v2/health/ready
+HEALTHCHECK --start-period=10m --interval=30s --retries=50 CMD curl --fail localhost:8501/v2/health/ready
 CMD [ "/models/start.py" ]
 
 FROM serving-develop AS serving-prod
@@ -38,7 +38,7 @@ RUN echo '[ -s "$NVM_DIR/bash_completion" ] && \. "$NVM_DIR/bash_completion"' >>
 RUN source /home/devuser/.bashrc && nvm install 20
 # Setup CI scripts
 COPY ./koina_*.sh /usr/local/bin/
-HEALTHCHECK --start-period=30s --interval=1m --retries=40 CMD [ "ls", "/tmp/done_setup" ]
+HEALTHCHECK --start-period=3m --interval=4m --retries=40 CMD [ "ls", "/tmp/done_setup" ]
 # Install pipx dependencies
 ARG PIPX_HOME=/home/devuser/.local/pipx/venvs
 ARG PIPX_BIN_DIR=/home/devuser/.local/bin

models/3dmolms/3dmolms_orbitrap_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/13928958/files/model_orbitrap.zip?download=1
-md5:0b3beb45d6ca6e7c82271bf3be9e3eee 
+md5:0b3beb45d6ca6e7c82271bf3be9e3eee 
+https://ndownloader.figshare.com/files/61421536

models/3dmolms/3dmolms_qtof_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/13928958/files/model_qtof.zip?download=1
-md5:8ac637326f9378c6c3d2535183ae439a
+md5:8ac637326f9378c6c3d2535183ae439a
+https://ndownloader.figshare.com/files/61458817

models/AlphaPept/AlphaPept_ccs_generic_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7576732/files/model.pt.zip?download=1
 md5:64bad0a110e8aa911520ce42c79c92c9
+https://ndownloader.figshare.com/files/61458970

models/AlphaPept/AlphaPept_ms2_generic_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7576749/files/model.pt.zip?download=1
 md5:07a99a478f284a58865785af90089a0a
+https://ndownloader.figshare.com/files/61459084

models/AlphaPept/AlphaPept_rt_generic_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7576813/files/model.pt.zip?download=1
 md5:27d0a2871dde5da1bff96f8733c359f9
+https://ndownloader.figshare.com/files/61459141

models/Altimeter/Altimeter_2024_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/14894501/files/model.pt.zip?download=1
 md5:714e2cd897171e42e61cce94bba28f8e
+https://ndownloader.figshare.com/files/61459279

models/Chronologer/Chronologer_RT_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/13742304/files/model.pt.zip?download=1
-md5:aa9f5fcdd7c074ca060e68799fdd764f 
+md5:aa9f5fcdd7c074ca060e68799fdd764f
+https://ndownloader.figshare.com/files/61459444

models/Deeplc/Deeplc_hela_hf_1f/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7971241/files/model.savedmodel.zip?download=1
 md5:831807c5938e52decb579a7968e4d11e
+https://ndownloader.figshare.com/files/61460860

models/Deeplc/IM2Deep_full_hc_peprec_CCS_v4_cb975cfdd4105f97efa0b3afffe075cc/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/13134482/files/model.savedmodel.zip?download=1
 md5:f29a01ed7af11a0f6df0670e1690aaf7
+https://ndownloader.figshare.com/files/61460983

models/MultiFrag/multifrag25/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/records/15064340/files/model.zip?download=1
 md5:329707c3cf59e2a56346656933fbf357
+https://ndownloader.figshare.com/files/61461658

models/Prosit/Prosit_2019_intensity_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7565518/files/model.savedmodel.zip?download=1
 md5:97a7493f4c48c8c4183ca00d3482056f
+https://ndownloader.figshare.com/files/61461964

models/Prosit/Prosit_2019_irt_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7671314/files/model.savedmodel.zip?download=1
 md5:4052b8ee575a7d92bc831f0fd1bbbbaa
+https://ndownloader.figshare.com/files/61462003

models/Prosit/Prosit_2019_irt_supplement_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7671350/files/model.savedmodel.zip?download=1
 md5:37a819bd15cf1a8ee017c90b6601afd6
+https://ndownloader.figshare.com/files/61470367

models/Prosit/Prosit_2020_intensity_CID_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7681435/files/model.savedmodel.zip?download=1
 md5:51fa15a10ab1ceb4bbb9e41e304ff5a5
+https://ndownloader.figshare.com/files/61462042

models/Prosit/Prosit_2020_intensity_HCD_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7681452/files/model.savedmodel.zip?download=1
 md5:0afd65f8bbe79cc792acb90f2ff611c1
+https://ndownloader.figshare.com/files/61465054

models/Prosit/Prosit_2020_intensity_TMT/notes.yaml

@@ -7,7 +7,7 @@ description: |
   Limitations of the training data: The model was not trained on peptides containing free cysteine side chains or other amino acid modifications that may be identified on HLA peptides. Also, the training data only included peptides with TMT label on the n-treminus.
   False positives/negatives: The training data was filtered for 1% FDR.
   **Optimization**
-  Optimization target: y and b fragment ion intensities with charge 1-3
+  Optimization target: y and b fragment ion intensities with charge 1-3.
   Experimental variability: No comparison with experimental variability is published.
   Metric: Spectral angle (SA) was used to compare predicted and experimental spectra.
   **Model**

models/Prosit/Prosit_2020_intensity_TMT_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7681498/files/model.savedmodel.zip?download=1
 md5:468235bd92d2a70bd1840feac6a4aab9
+https://ndownloader.figshare.com/files/61465057

models/Prosit/Prosit_2020_irt_TMT/notes.yaml

@@ -7,9 +7,9 @@ description: |
   Limitations of the training data: The model was trained on tryptic peptides from human origin. There is a possibility that additional liquid chromatography parameters not captured by iRT values may lead to suboptimal predictions.
   False positives/negatives: The training data was filtered for 1% FDR.
   **Optimization**
-  Optimization target: y and b fragment ion intensities with charge 1-3
+  Optimization target: iRT values for peptides.
   Experimental variability: No comparison with experimental variability is published.
-  Metric: Spectral angle (SA) was used to compare predicted and experimental spectra.
+  Metric: Mean Absolute Error (MAE) was used to compare predicted and experimental iRT.
   **Model**
   Interpretability: The model is a black box, However, the model is split into an encoder and a decoder, which means the model learns an intermediate latent space to represent peptide sequences and parameters which is then used to make predictions.
   Model type: The model performs regression, predicting a continuous iRT value.

models/Prosit/Prosit_2020_irt_TMT_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/7681553/files/model.savedmodel.zip?download=1
 md5:425586cf82f75a049d330cc143fac8f1
+https://ndownloader.figshare.com/files/61465060

models/Prosit/Prosit_2023_intensity_timsTOF_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/8211811/files/model.savedmodel.zip?download=1
 md5:cc476156badc4bee501c42568bd68b15
+https://ndownloader.figshare.com/files/61465069

models/Prosit/Prosit_2024_intensity_PTMs_gl_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/14060540/files/model.savedmodel.zip?download=1
-md5:6d235a40dc613c5dcb170e7ea6ca9b84
+md5:6d235a40dc613c5dcb170e7ea6ca9b84
+https://ndownloader.figshare.com/files/61465078

models/Prosit/Prosit_2024_intensity_cit_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/14673884/files/model.savedmodel.zip?download=1
-md5:9e3a2cdca019767ded49eeb6efe8e6a3
+md5:9e3a2cdca019767ded49eeb6efe8e6a3
+https://ndownloader.figshare.com/files/61465075

models/Prosit/Prosit_2024_irt_PTMs_gl_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/14060493/files/model.savedmodel.zip?download=1
-md5:0f9196ee4b4cf381de476afccb38c61e
+md5:0f9196ee4b4cf381de476afccb38c61e
+https://ndownloader.figshare.com/files/61465084

models/Prosit/Prosit_2024_irt_cit_core/1/.zenodo

@@ -1,2 +1,3 @@
 https://zenodo.org/record/13891386/files/model.savedmodel.zip?download=1
-md5:d97e9c5b1993e9a8121f8dd8031dffaa
+md5:d97e9c5b1993e9a8121f8dd8031dffaa
+https://ndownloader.figshare.com/files/61465081

models/Prosit/Prosit_2025_intensity_22PTM/config.pbtxt

@@ -0,0 +1,212 @@
+max_batch_size: 1000
+platform: "ensemble"
+input [
+  {
+   name: 'peptide_sequences',
+   data_type: TYPE_STRING,
+   dims: [-1]
+  },
+  {
+    name: 'precursor_charges',
+    data_type: TYPE_INT32,
+    dims: [1],
+  },
+  {
+    name: 'collision_energies',
+    data_type: TYPE_FP32,
+    dims: [1],
+  },
+  {
+    name: 'fragmentation_types',
+    data_type: TYPE_STRING,
+    dims: [1],
+  }
+]
+output [
+  {
+   name: 'intensities',
+   data_type: TYPE_FP32,
+   dims: [174]
+  },
+ {
+   name: 'mz',
+   data_type: TYPE_FP32,
+   dims: [174]
+ },
+ {
+   name: 'annotation',
+   data_type: TYPE_STRING,
+   dims: [174]
+ }
+]
+
+ensemble_scheduling {
+  step [
+    {
+      model_name: "Prosit_Preprocess_charge_cit"
+      model_version: 1
+      input_map {
+        key: "precursor_charges"
+        value: "precursor_charges"
+      },
+      output_map {
+        key: "precursor_charges_in:0"
+        value: "precursor_charges_in_preprocessed:0"
+      }
+    },
+	{
+      model_name: "Prosit_Preprocess_charge"
+      model_version: 1
+      input_map {
+        key: "precursor_charges"
+        value: "precursor_charges"
+      },
+      output_map {
+        key: "precursor_charges_in:0"
+        value: "precursor_charges_in_preprocessed_FP"
+      }
+    },
+	{
+      model_name: "Prosit_Preprocess_unmod"
+      model_version: 1
+       input_map {
+        key: "peptide_sequences"
+        value: "peptide_sequences"
+      },
+      output_map {
+        key: "stripped_peptide"
+        value: "stripped_peptide"
+      }
+    },
+	{
+      model_name: "Prosit_Preprocess_ac_gain"
+      model_version: 1
+       input_map {
+        key: "peptide_sequences"
+        value: "peptide_sequences"
+      },
+      output_map {
+        key: "ac_gain"
+        value: "ac_gain"
+      }
+    },
+	{
+      model_name: "Prosit_Preprocess_ac_loss"
+      model_version: 1
+       input_map {
+        key: "peptide_sequences"
+        value: "peptide_sequences"
+      },
+      output_map {
+        key: "ac_loss"
+        value: "ac_loss"
+      }
+    },
+    {
+      model_name: "Prosit_Preprocess_mod_peptide"
+      model_version: 1
+      input_map {
+        key: "peptide_sequences"
+        value: "stripped_peptide"
+      },
+      output_map {
+        key: "peptides_in:0"
+        value: "peptides_in:0"
+      }
+    },
+    {
+      model_name: "Prosit_Preprocess_collision_energy"
+      model_version: 1
+      input_map {
+        key: "raw_collision_energy"
+        value: "collision_energies"
+      },
+      output_map {
+        key: "norm_collision_energy"
+        value: "norm_collision_energy"
+      }
+    },
+	 {
+      model_name: "Prosit_Preprocess_fragmentation_types_PTMs"
+      model_version: 1
+      input_map {
+        key: "fragmentation_types"
+        value: "fragmentation_types"
+      },
+      output_map {
+        key: "fragmentation_types_encoding"
+        value: "fragmentation_types_encoding"
+      }
+    },
+    {
+      model_name: "Prosit_2025_intensity_22PTM_core"
+      model_version: 1
+      input_map {
+        key: "modified_sequence"
+        value: "peptides_in:0"
+      },
+	  input_map {
+        key: "mod_loss"
+        value: "ac_loss"
+      },
+	  input_map {
+        key: "mod_gain"
+        value: "ac_gain"
+      },
+      input_map {
+        key: "collision_energy_aligned_normed"
+        value: "norm_collision_energy"
+      },
+	  input_map {
+        key: "method_nbr"
+        value: "fragmentation_types_encoding"
+      },
+
+      input_map {
+        key: "precursor_charge_onehot"
+        value: "precursor_charges_in_preprocessed:0"
+      }
+      output_map {
+        key: "output_1"
+        value: "output_1"
+      }
+    },
+
+    {
+      model_name: "Prosit_2019_intensity_postprocess"
+      model_version: 1
+      input_map {
+        key: "peptides_in:0"
+        value: "peptide_sequences"
+      },
+      input_map{
+        key: "precursor_charges_in:0"
+        value: "precursor_charges_in_preprocessed_FP"
+      }
+      input_map{
+        key: "peaks_in:0",
+        value: "output_1"
+      }
+      output_map {
+        key: "intensities"
+        value: "intensities"
+      }
+      output_map {
+        key: "mz"
+        value: "mz"
+      }
+    },
+    {
+      model_name: "Prosit_Helper_annotation"
+      model_version: 1
+      input_map {
+        key: "precursor_charges"
+        value: "precursor_charges"
+      },
+      output_map {
+        key: "annotation"
+        value: "annotation"
+      }
+    }
+  ]
+}