Skip to content

Commit

Permalink
Merge pull request #2 from taylor-lab/python3
Browse files Browse the repository at this point in the history 
Support for Python 3
  • Loading branch information
@cband
cband authored Jul 26, 2019
2 parents a873fc9 + c34d7c5 commit c3c3c75
Show file tree
Hide file tree
Showing 4 changed files with 95 additions and 88 deletions.
6 changes: 5 additions & 1 deletion .travis.yml
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,10 @@
language: python
dist: xenial # required for Python >= 3.7
python:
- 2.7
- "2.7"
- "3.5"
- "3.6"
- "3.7"
install:
- sudo apt-get -y update
- pip install codecov
Expand Down
62 changes: 32 additions & 30 deletions README.md
View file
Original file line number Diff line number Diff line change
@@ -1,29 +1,28 @@


# neoantigen-dev

MHC Class I neoantigen prediction pipeline from IM5/IM6/WES/WGS. Takes Normal `.bam` and somatic `.maf` and generates neoantigen predictions for HLA-A/B/C.

The pipeline has four main steps:

1. **Genotype HLA**. genotyping performed using POLYSOLVER.
2. **Construct mutated peptides**. For non-synonymous mutations, generates mutated peptide sequences based on `HGVSc`. _NOTE_: `.maf` file should be VEP annotated using `cmo_maf2maf --version 1.6.14 --vep-release 88` **using this EXACT VERSION**. TODO: Generate mutated sequences for fusions.
3. **Run NetMHCpan-4.0 and NetMHC-4.0**. using default parameters for each algorithm.
1. **Genotype HLA**. genotyping performed using POLYSOLVER.
2. **Construct mutated peptides**. For non-synonymous mutations, generates mutated peptide sequences based on `HGVSc`. _NOTE_: `.maf` file should be VEP annotated using `cmo_maf2maf --version 1.6.14 --vep-release 88` **using this EXACT VERSION**. TODO: Generate mutated sequences for fusions.
3. **Run NetMHCpan-4.0 and NetMHC-4.0**. using default parameters for each algorithm.
4. **Post-processing**. compiles predictions from both algorithms and finds strongest binder for each non-synonymous mutation. Also, each predicted neopeptide is searched against the entire reference peptidome to make sure it is a true neopeptide. `is_in_wt_peptidome` column reflects that. TODO: Incorporate neoantigen quality from [Lukzsa et al., Nature 2017](https://www.nature.com/articles/nature24473)


## Install

Clone the repo and install any necessary python2 libraries from `requirements.txt`. Note that this repo is currently only compatible with Python 2.7, not Python 3.x :
Clone the repo and install any necessary Python libraries from `requirements.txt`. This repo is currently compatible with Python 2 and 3. To install:

```bash
git clone https://github.com/taylor-lab/neoantigen-dev.git
cd neoantigen-dev
pip install -r requirements.txt
```


## Usage
NOTE: For POLYSOLVER step, the pipeline requires 8 cores.

NOTE: For POLYSOLVER step, the pipeline requires 8 cores.

```
# Neoantigen prediction pipeline. Four main steps:
(1) Genotype HLA using POLYSOLVER,
Expand Down Expand Up @@ -66,44 +65,48 @@ Optional arguments:
--force_rerun_netmhc ignores any existing netMHCpan output and re-runs it.
Default: false
```

## Output

### HLA genotypes

```
<output_dir>/polysolver/winners.hla.txt
```

### Neoantigen binding affinties annotated MAF

```
<sample_id>.neoantigens.maf. (peptide with the highest binding affinity is incorporated into the original .maf for each non-syn mutation)
<sample_id>.neoantigens.maf. (peptide with the highest binding affinity is incorporated into the original .maf for each non-syn mutation)
<sample_id>.all_neoantigen_predictions.txt: all the predictions made for all peptides by both the algorithms
```

The following columns are appended to the input `.maf`.

| Column Name | Description |
| ------------- |:-------------|
| neo_maf_identifier_key | a unique key that can be used to find other peptides predicted for the same mutation (in `.all_neoantigen_predictions.txt`) |
| neo_best_icore_peptide | neopeptide sequence for the strongest binder |
| neo_best_rank | binding rank for the strongest binder |
| neo_best_binding_affinity | binding affinity for the strongest binder |
| neo_best_binder_classification | binding classification for the strongest binder (`Non Binder`, `Strong Binder`, `Weak Binder`) |
| neo_best_is_in_wt_peptidome | `TRUE`/`FALSE` indicating whether the strongest binder peptide is in the reference peptidome |
| neo_best_algorithm | algorithm predicting the strongest binder |
| neo_best_hla_allele | hla allele for the strongest binder |
| neo_n_peptides_evaluated | total # of all peptides evaluated (unique icore peptides) |
| neo_n_strong_binders | total # of strong binders |
| neo_n_weak_binders | total # of weak binders |
| Column Name | Description |
| ------------------------------ | :-------------------------------------------------------------------------------------------------------------------------- |
| neo_maf_identifier_key | a unique key that can be used to find other peptides predicted for the same mutation (in `.all_neoantigen_predictions.txt`) |
| neo_best_icore_peptide | neopeptide sequence for the strongest binder |
| neo_best_rank | binding rank for the strongest binder |
| neo_best_binding_affinity | binding affinity for the strongest binder |
| neo_best_binder_classification | binding classification for the strongest binder (`Non Binder`, `Strong Binder`, `Weak Binder`) |
| neo_best_is_in_wt_peptidome | `TRUE`/`FALSE` indicating whether the strongest binder peptide is in the reference peptidome |
| neo_best_algorithm | algorithm predicting the strongest binder |
| neo_best_hla_allele | hla allele for the strongest binder |
| neo_n_peptides_evaluated | total # of all peptides evaluated (unique icore peptides) |
| neo_n_strong_binders | total # of strong binders |
| neo_n_weak_binders | total # of weak binders |

The column description for `.all_neoantigen_predictions.txt` can be found in: [http://www.cbs.dtu.dk/services/NetMHC/output.php](http://www.cbs.dtu.dk/services/NetMHC/output.php). Additional columns are:

The following columns are appended to the input `.maf`.

| Column Name | Description |
| ------------- |:-------------|
| binder_class | `Non Binder`, `Strong Binder` (`rank < 0.5 or affinity < 50`), `Weak Binder` (`rank < 2 or affinity < 500`) |
| best_binder_for_icore_group | `TRUE`/`FALSE` indicating if the binding prediction is the strongest among all the HLA-alleles/algorithms for the given icore peptide. |
| is_in_wt_peptidome | if the peptide is present in any other protein in the entire peptidome |
| neo_maf_identifier_key | a unique key that can be used to find other peptides predicted for the same mutation (in `.neoantigens.maf`) |
| Column Name | Description |
| --------------------------- | :------------------------------------------------------------------------------------------------------------------------------------- |
| binder_class | `Non Binder`, `Strong Binder` (`rank < 0.5 or affinity < 50`), `Weak Binder` (`rank < 2 or affinity < 500`) |
| best_binder_for_icore_group | `TRUE`/`FALSE` indicating if the binding prediction is the strongest among all the HLA-alleles/algorithms for the given icore peptide. |
| is_in_wt_peptidome | if the peptide is present in any other protein in the entire peptidome |
| neo_maf_identifier_key | a unique key that can be used to find other peptides predicted for the same mutation (in `.neoantigens.maf`) |

## Example

Expand All @@ -114,4 +117,3 @@ python neoantigen.py --config_file neoantigen-luna.config \
--output_dir <output_dir> \
--maf_file <cmo_vep_annotated_maf_file>
```

Loading

0 comments on commit c3c3c75

Please sign in to comment.