The chill-filter user guide

Home page | Frequently Asked questions

chill-filter is a rapid sample-screening Web application that comprehensively determines the composition of a shotgun sequence data set. It should take about 10 seconds to process an uploaded sample.

You could use it to:

• detect microbial contaminants in a eukaryotic genome sequencing project (e.g. Bu5);
• detect host contamination in a microbiome sequencing project (e.g. SRR5650070);
• figure out what's in an unknown or problematic sample (e.g. ERR2245457);

chill-filter will sensitively detect the presence of large genomes (500kb and up) at sequencing depths as low as 0.1x. It searches your entire sample: it does not select a subset of your data.

The main value of chill-filter is that it is fast, sensitive, and comprehensive. Within the limits of the available reference database, it should detect the presence of all microbial and eukaryotic genomes.

Caveats: If you want a full taxonomic breakdown of your sample, there are other tools you might use, e.g. sourmash, but they don't run in a Web browser :). You can't currently look for viruses or other small genomes, but we would be happy to talk about how to do that. Chill-filter also does not work on amplicon data sets, and the numbers need to be treated with skepticism when analyzing enrichment data sets (ChIP-seq or ATAC-seq) or RNA data.

chill-filter is a research service and should not be relied upon for clinical diagnostics or confirmatory analyses.

What are chill-filter's goals?

chill-filter aims to be a sensitive, specific, and comprehensive approach to figuring out what is in your sample. It will always underestimate the total known content in your sample.

chill-filter will never share your sequence data (we don't actually have access to it, anyway!) However, please don't upload sensitive or personally identifying data.

How do you use chill-filter?

On the front page, select a FASTA or FASTQ file to upload. The file can be in a compressed (gzip or bzip2) format, and can be either raw sequencing data (e.g. Illumina, PacBio, or Nanopore) or an assembly. No prior filtering or processing of the data is required, and we don't recommend doing any, since it may affect interpretation of the results.

This file will be processed and compressed locally, in your browser, and then uploaded to the chill-filter Web site. Most data sets will compress to under 1 MB.

After a brief delay (hopefully under 10 seconds!) you should see a report of what chill-filter has detected in your data!

If given reads, chill-filter will estimate the abundances of the detected genomes in the read data set, as well as the total amount of sequence (in reads) that should map. See the Bu5 read data set for an example.

If given an assembly, chill-filter will estimate the total number of bases that should align between the assembly and the detected genomes. See the Bu5 assembly data set for an example.

chill-filter estimates are generally a significant underestimate of mapping rates and abundances. This is largely a result of differences between what is being sequenced and what is in the database; chill-filter cannot detect or account for these differences.

What does chill-filter look for?

As of Dec 2024, chill-filter matches against the human genome, 8 individual animal genomes (cow, dog, horse, cat, chicken, mouse, goat, and pig), and two combined databases - GenBank plants and GTDB R220 microbes. These last two databases are merged, so you can only detect that there is a match, without seeing what the matching genomes are.

GenBank plants contains approximately 1700 plant genomes (all plant genomes as of Dec 2024). GTDB RS220 contains approximately 584,000 bacterial genomes and 12,500 archaeal genomes.

Questions, comments, or requests?

Please visit the Frequently Asked Questions page, or ask a question in the issue tracker! You can also e-mail us at chill@sourmash.bio.

Who is responsible for chill-filter?

chill-filter is a product of the sourmash project. It is currently unfunded, although the sourmash software stack it is built upon has been funded in part by the Moore Foundation, the NIH, the USDA, and the NSF.