The ITSNdb is a new neoantigen database with know immunogenic and non immunogenic tumor specific antigenic peptides derived from genomic rearrangements, such as single nucleotide variants (SNVs), that satisfy the following criteria:
- The wild type counterpart has been identified in the source protein
- The MHC-I presentation has been experimentally validated
- The positive or negative immunogenicity has been experimentally validated by, for instance, ELISPOT®
In this sence, all peptides in the database have experimental confirmation of their positive/negative immunogenicity (classified as “Positive” and “Negative” neoantigens respectively) as well as their cell surface presentation.The neoantigens were collected and curated from published articles searched on PubMedTM using “neoantigen'' or “neoepitopes” as keywords. The ITSNdb only includes neoantigens whose inclusion criteria were explicitly described in its reference bibliography.
ITSNdb APP is now available on line!!
Nibeyro et al. Unraveling Tumor Specific Neoantigen immunogenicity prediction: a comprehensive analysis Front. Immunol.Sec. Cancer Immunity and Immunotherapy Volume 14 - 2023 | doi: 10.3389/fimmu.2023.1094236
Nibeyro et al. MHC-I binding affinity derived metrics fail to predict tumor specific neoantigen immunogenicity BioRxiv
In order to install the ITSNdb R library the following tools are required:
- The Multiple Sequence Alignment (msa) R library available on Bioconductor
install.packages("remotes")
library(remotes)
install.packages("devtools")
library(devtools)
install_github("elmerfer/ITSNdb")
##load library
library(ITSNdb)
##load the data
data(ITSNdb)
-
Val_dataset: validation dataset that simulates a patient neoantigen landscape, which includes 113 non-immunogenic neopeptides-HLA pairs with unvalidated MHC-I presentation; and 7 immunogenic, non-SNV derived, neoantigens-HLA pairs with both MHC-I presentation and immunogenicity experimentally validated.
- References: Robbins, P. et al., Ehx, G. et al., Huang, J. et al. and Yang, W. et al..
- Main usage: performance validation, prioritization evaluation.
-
TNB_dataset: dataset containing a list of candidates tumor specific neoantigens predicted to bind to the MHC-I complex, of patients from six immune checkpoint blockade immunotherapy (ICB) treated cohorts, one non–small cell lung and five melanoma cancer cohorts, with ICB response association and TMB evaluation.
- References: Rizvi, N. A. et al., Van Allen, E. M. et al., Snyder, A. et al., Riaz, N. et al..
- Main usage: biomarker applicability evaluation.
Once ITSNdb installed
## Load library
library(ITSNdb)
## Load the data
data(Val_dataset)
data(TNB_dataset)Here, in order to facilitate the exploration of state of the art tools for binding affinity prediction or immunogenicity score prediction, we implement easy to use interfaces for peptide-HLA binding affinity or immunogenicity prediction:
- Through R: netMHCpan, PRIME & mixMHCpred and The Class I Immunogenicity IEDB predictor
- Through Colab : MHCflurry, DeepImmune and DeepHLApan.
In all cases the same file can be used to feed any platform, thus allowing easy comparison of the different methods. The R and Colab interfaces were implemented to facilitate the analisis of subject specific peptide-HLA pair list.
Data Format See Sample
In order to feed the methods, the file should contain peptide-HLA pairs with the following format
The ITSNdb library allows the installation and use of the netMHCpan version 4.1 software to predict peptide binding affinity to MHC-I molecules. (up to now only available for Linux, Mac in progress)
To verify if you have it in your machine, please type from a console terminal the following command
'tcsh --version'
if succeed you will see something like this:
if not installed try 'sudo apt-get install tcsh' and verify.
Follow the instructions and fill the form to receive the rights to download netMHCpan
and save it in your favorite directory.
Onpen an R session or RStudio and type:
install.packages("devtools")
library(devtools)
install_github("elmerfer/ITSNdb")
#load library
library(ITSNdb)
#file.choose() will open a window selector to look for the downloaded file
Install_netMHCPan(file.choose(), dir = "/where i whant to install it/dir")
#if success, the following message should appear
netMHCpan Installation OKdata(ITSNdb)
ITSNdb$Neoantigen[1]
[1] "GRIAFFLKY"
ITSNdb$HLA[1:2]
[1] "HLA-B27:05" "HLA-B35:03"
results <- RunNetMHCPan_peptides(peps=ITSNdb$Neoantigen[1], alleles = ITSNdb$HLA[1:2])
results
$`HLA-B27:05`
Pos MHC Peptide Core Of Gp Gl Ip Il Icore Identity Score_EL %Rank_EL Score_BA %Rank_BA Aff(nM) BindLevel
1 1 HLA-B*27:05 GRIAFFLKY GRIAFFLKY 0 0 0 0 0 GRIAFFLKY PEPLIST 0.9810120 0.009 0.659218 0.103 39.93 SB
$`HLA-B35:03`
Pos MHC Peptide Core Of Gp Gl Ip Il Icore Identity Score_EL %Rank_EL Score_BA %Rank_BA Aff(nM) BindLevel
1 1 HLA-B*35:03 GRIAFFLKY GRIAFFLKY 0 0 0 0 0 GRIAFFLKY PEPLIST 0.0000600 31.333 0.009315 55.345 45206.31 <NA>## we will build a simulated cohort study with two subjects
df.to.test <- data.frame(Sample = c("Subject1","Subject1","Subject2"), Neoantigen=ITSNdb$Neoantigen[1:3],HLA = ITSNdb$HLA[1:3])
df.to.test
Sample Neoantigen HLA
1 Subject1 GRIAFFLKY HLA-B27:05
2 Subject1 LPIQYEPVL HLA-B35:03
3 Subject2 KLILWRGLK HLA-A03:01
# we will save it in a comma separated text file in the working directory
write.csv(df.to.test,file="MyPatientsNeoantigenList.csv",quote=F, row.names = F)
#run predictions
Cohort_results <- RunNetMHCPan(pepFile = "MyPatientsNeoantigenList.csv")
Cohort_results
Neoantigen Sample HLA NetMCpan_Pos NetMCpan_MHC NetMCpan_Core NetMCpan_Of NetMCpan_Gp NetMCpan_Gl NetMCpan_Ip NetMCpan_Il
1 GRIAFFLKY Subject1 HLA-B27:05 1 HLA-B*27:05 GRIAFFLKY 0 0 0 0 0
2 LPIQYEPVL Subject1 HLA-B35:03 1 HLA-B*35:03 LPIQYEPVL 0 0 0 0 0
3 KLILWRGLK Subject2 HLA-A03:01 1 HLA-A*03:01 KLILWRGLK 0 0 0 0 0
NetMCpan_Icore NetMCpan_Identity NetMCpan_Score_EL NetMCpan_%Rank_EL NetMCpan_Score_BA NetMCpan_%Rank_BA NetMCpan_Aff(nM)
1 GRIAFFLKY PEPLIST 0.9810120 0.009 0.659218 0.103 39.93
2 LPIQYEPVL PEPLIST 0.9816510 0.004 0.690369 0.008 28.51
3 KLILWRGLK PEPLIST 0.7217850 0.184 0.733193 0.049 17.94
NetMCpan_BindLevel
1 SB
2 SB
3 SBInstallation of PRIME (PRedictor of class I IMmunogenic Epitopes) in R using the ITSNdb
PRIME is a PRedictor of class I IMmunogenic Epitopes. It combines predictions of binding to HLA-I molecules and propensity for TCR recognition.
Here we provide an R interface to install and use PRIME in your local machine.
Onpen an R session or RStudio and type:
install.packages("devtools")
library(devtools)
install_github("elmerfer/ITSNdb")
#load library
library(ITSNdb)
Install_PRIME(dir = "/where i whant to install it/dir")
#if success, the following message should appear
PRIME Installation OK## we will build a simulated cohort study with two subjects
df.to.test <- data.frame(Sample = c("Subject1","Subject1","Subject2"), Neoantigen=ITSNdb$Neoantigen[1:3],HLA = ITSNdb$HLA[1:3])
df.to.test
Sample Neoantigen HLA
1 Subject1 GRIAFFLKY HLA-B27:05
2 Subject1 LPIQYEPVL HLA-B35:03
3 Subject2 KLILWRGLK HLA-A03:01
# we will save it in a comma separated text file in the working directory
write.csv(df.to.test,file="MyPatientsNeoantigenList.csv",quote=F, row.names = F)
#run predictions
Cohort_results <- RunPRIME(pepFile = "MyPatientsNeoantigenList.csv")
Cohort_results
Neoantigen Sample HLA PRIME_Rank_bestAllele PRIME_Score_bestAllele PRIME_RankBinding_bestAllele PRIME_BestAllele PRIME_Rank
1 GRIAFFLKY Subject1 HLA-B27:05 0.001 0.302405 0.012 B2705 0.001
2 LPIQYEPVL Subject1 HLA-B35:03 0.001 0.312395 0.001 B3503 0.001
3 KLILWRGLK Subject2 HLA-A03:01 0.132 0.108694 0.336 A0301 0.132
PRIME_Score PRIME_RankBinding
1 0.302405 0.012
2 0.312395 0.001
3 0.108694 0.336You can also predict only binding affinity scores trough mixMHCpred (automatically installed when installing PRIME)
## we will build a simulated cohort study with two subjects
df.to.test <- data.frame(Sample = c("Subject1","Subject1","Subject2"), Neoantigen=ITSNdb$Neoantigen[1:3],HLA = ITSNdb$HLA[1:3])
df.to.test
Sample Neoantigen HLA
1 Subject1 GRIAFFLKY HLA-B27:05
2 Subject1 LPIQYEPVL HLA-B35:03
3 Subject2 KLILWRGLK HLA-A03:01
# we will save it in a comma separated text file in the working directory
write.csv(df.to.test,file="MyPatientsNeoantigenList.csv",quote=F, row.names = F)
#run predictions
Cohort_results <- RunMixMHCpred(pepFile = "MyPatientsNeoantigenList.csv")
Cohort_results
Neoantigen Sample HLA MixMHCpred_Score_bestAllele MixMHCpred_BestAllele MixMHCpred_Rank_bestAllele MixMHCpred_Score
1 GRIAFFLKY Subject1 HLA-B27:05 0.498937 B2705 0.0118114 0.498937
2 LPIQYEPVL Subject1 HLA-B35:03 1.243759 B3503 0.0010000 1.243759
3 KLILWRGLK Subject2 HLA-A03:01 -0.428836 A0301 0.3362680 -0.428836
MixMHCpred_Rank
1 0.0118114
2 0.0010000
3 0.3362680Estimating immunogenicity score by means of the Class I Immunogenicity IEDB function developped by Calis et al
## we will build a simulated cohort study with two subjects
df.to.test <- data.frame(Sample = c("Subject1","Subject1","Subject2"), Neoantigen=ITSNdb$Neoantigen[1:3],HLA = ITSNdb$HLA[1:3])
df.to.test
Sample Neoantigen HLA
1 Subject1 GRIAFFLKY HLA-B27:05
2 Subject1 LPIQYEPVL HLA-B35:03
3 Subject2 KLILWRGLK HLA-A03:01
# we will save it in a comma separated text file in the working directory
write.csv(df.to.test,file="MyPatientsNeoantigenList.csv",quote=F, row.names = F)
#run predictions
Cohort_results <- RunClassIImmunogenicity(pepFile = "MyPatientsNeoantigenList.csv")
Cohort_results
Sample Neoantigen HLA CIImm
1 Subject1 GRIAFFLKY HLA-B27:05 0.17141
2 Subject1 LPIQYEPVL HLA-B35:03 0.03205
3 Subject2 KLILWRGLK HLA-A03:01 0.31858Estimate immunogenic scores or affinities of peptide-HLA pairs from ITSNdb or cohort studies by using DeepHLApan, DeepImmune and MHCflurry in Colab environments
Each colab session allows to upload only one file, to upload a second one please terminate the current session and start a new one
