Reverse-complement parameter sharing improves deep learning models for genomics

Avanti Shrikumar; Peyton Greenside; Anshul Kundaje

doi:10.1101/103663

CARVIEW

MOTORHOMES

Select Language

HTTP/2 301 date: Mon, 19 Jan 2026 15:58:08 GMT content-length: 0 location: https://doi.org/10.1101/103663 server: cloudflare vary: Origin expires: Tue, 20 Jan 2026 15:58:08 GMT permissions-policy: interest-cohort=(),browsing-topics=() cf-cache-status: DYNAMIC nel: {"report_to":"cf-nel","success_fraction":0.0,"max_age":604800} strict-transport-security: max-age=31536000; includeSubDomains; preload report-to: {"group":"cf-nel","max_age":604800,"endpoints":[{"url":"https://a.nel.cloudflare.com/report/v4?s=q9deHbLpIe%2BpzO2l4ihW37tl4DVr%2FAunQuxPlsQYo%2FALgZkCDEyDcYaY9rutRW%2BJB9HtWYBZu8GrkD3RRQ3HMaCnRVJfDQ%3D%3D"}]} cf-ray: 9c0788297816cb77-BLR alt-svc: h3=":443"; ma=86400 HTTP/2 302 date: Mon, 19 Jan 2026 15:58:09 GMT content-type: text/html;charset=utf-8 location: https://biorxiv.org/lookup/doi/10.1101/103663 server: cloudflare vary: Origin vary: Accept expires: Mon, 19 Jan 2026 16:11:00 GMT permissions-policy: interest-cohort=(),browsing-topics=() cf-cache-status: DYNAMIC nel: {"report_to":"cf-nel","success_fraction":0.0,"max_age":604800} strict-transport-security: max-age=31536000; includeSubDomains; preload report-to: {"group":"cf-nel","max_age":604800,"endpoints":[{"url":"https://a.nel.cloudflare.com/report/v4?s=pgGI5uQc3B0YrnF5kKuH7gWbbBc7zluWbmuwpgR5LxswJZfVCDqDKtpepbw08ZRuGwCHLy%2Fr6onTb57nsZWX5hQJ9b5vhQ%3D%3D"}]} cf-ray: 9c078829c89bcb77-BLR alt-svc: h3=":443"; ma=86400 HTTP/1.1 302 Found Date: Mon, 19 Jan 2026 15:58:09 GMT Content-Type: text/html; charset=iso-8859-1 Transfer-Encoding: chunked Connection: keep-alive server: cloudflare location: https://www.biorxiv.org/lookup/doi/10.1101/103663 cf-cache-status: DYNAMIC Nel: {"report_to":"cf-nel","success_fraction":0.0,"max_age":604800} Report-To: {"group":"cf-nel","max_age":604800,"endpoints":[{"url":"https://a.nel.cloudflare.com/report/v4?s=vVUN%2B0MCcGGx5qsOwHs95QX77BoL0yOEYM44jMaMCyO8XGFtDdJCCy5CI9wuyrxMtLnOwiSg2cJAOaEEb%2B6%2BqJwFapgPTxi9lA%3D%3D"}]} CF-RAY: 9c07882b69f431be-BOM alt-svc: h3=":443"; ma=86400 HTTP/2 301 date: Mon, 19 Jan 2026 15:58:10 GMT content-type: text/html; charset=UTF-8 location: https://www.biorxiv.org/content/10.1101/103663v1 cf-ray: 9c07882e9a9cc188-BLR x-content-type-options: nosniff x-content-type-options: nosniff x-drupal-cache: MISS expires: Mon, 19 Jan 2026 16:28:10 GMT cache-control: public, max-age=1800 x-varnish-ttl: pragma: no-cache vary: Accept-Encoding x-highwire-sitecode: biorxiv x-highwire-smart-code: biorxiv_production x-varnish: 1901630920 x-varnish-cache: via: 1.1 varnish cf-cache-status: EXPIRED set-cookie: __cf_bm=WIOvl4O96WrsMsecHlRwe3pBvM9CNJz9n09EHFOJ_6k-1768838290-1.0.1.1-UC_32wT1is1nZ7wbnkfaqBg2Q7YT9pHeAwkzRmPOHa2bhTZeJonPf5B5GE75aAF.JmNHaOYA_WRvTDUVKD.ktQn_GxekpIPut3HySMQo7CQ; path=/; expires=Mon, 19-Jan-26 16:28:10 GMT; domain=.www.biorxiv.org; HttpOnly; Secure; SameSite=None server: cloudflare HTTP/2 200 date: Mon, 19 Jan 2026 15:58:11 GMT content-type: text/html; charset=utf-8 content-encoding: gzip x-content-type-options: nosniff x-content-type-options: nosniff x-drupal-cache: MISS expires: Sun, 19 Nov 1978 05:00:00 GMT cache-control: no-cache, must-revalidate set-cookie: SSESS1dd6867f1a1b90340f573dcdef3076bc=v08TxXS1VkKqD5cgpBLNAwMHqmKz7pbpCGVcjFYHZGM; expires=Wed, 11-Feb-2026 19:31:30 GMT; path=/; domain=.biorxiv.org; secure; HttpOnly content-language: en x-frame-options: SAMEORIGIN x-generator: Drupal 7 (https://drupal.org) link: ; rel="canonical",; rel="shortlink" vary: Accept-Encoding x-highwire-sitecode: biorxiv x-highwire-smart-code: biorxiv_production x-varnish: 1901630952 age: 0 via: 1.1 varnish x-varnish-ttl: x-varnish-cache: cf-cache-status: DYNAMIC server: cloudflare cf-ray: 9c0788326d1bc188-BLR Reverse-complement parameter sharing improves deep learning models for genomics | bioRxiv

Abstract

Deep learning approaches that have produced breakthrough predictive models in computer vision, speech recognition and machine translation are now being successfully applied to problems in regulatory genomics. However, deep learning architectures used thus far in genomics are often directly ported from computer vision and natural language processing applications with few, if any, domain-specific modifications. In double-stranded DNA, the same pattern may appear identically on one strand and its reverse complement due to complementary base pairing. Here, we show that conventional deep learning models that do not explicitly model this property can produce substantially different predictions on forward and reverse-complement versions of the same DNA sequence. We present four new convolutional neural network layers that leverage the reverse-complement property of genomic DNA sequence by sharing parameters between forward and reverse-complement representations in the model. These layers guarantee that forward and reverse-complement sequences produce identical predictions within numerical precision. Using experiments on simulated and in vivo transcription factor binding data, we show that our proposed architectures lead to improved performance, faster learning and cleaner internal representations compared to conventional architectures trained on the same data.

Availability Our implementation is available at

Contact avanti{at}stanford.edu, pgreens{at}stanford.edu, akundaje{at}stanford.edu